U.S. patent application number 16/854329 was filed with the patent office on 2021-05-27 for personalized tailored air interface.
This patent application is currently assigned to HUAWEI TECHNOLOGIES CO., LTD.. The applicant listed for this patent is MING JIA, JIANGLEI MA, WEN TONG, PEIYING ZHU. Invention is credited to MING JIA, JIANGLEI MA, WEN TONG, PEIYING ZHU.
Application Number | 20210160149 16/854329 |
Document ID | / |
Family ID | 1000004793694 |
Filed Date | 2021-05-27 |
View All Diagrams
United States Patent
Application |
20210160149 |
Kind Code |
A1 |
MA; JIANGLEI ; et
al. |
May 27, 2021 |
PERSONALIZED TAILORED AIR INTERFACE
Abstract
Methods and devices utilizing artificial intelligence (AI) or
machine learning (ML) for customization of a device specific air
interface configuration in a wireless communication network are
provided. An over the air information exchange to facilitate the
training of one or more AI/ML modules involves the exchange of
AI/ML capability information identifying whether a device supports
AI/ML for optimization of the air interface.
Inventors: |
MA; JIANGLEI; (OTTAWA,
CA) ; ZHU; PEIYING; (OTTAWA, CA) ; TONG;
WEN; (OTTAWA, CA) ; JIA; MING; (OTTAWA,
CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
MA; JIANGLEI
ZHU; PEIYING
TONG; WEN
JIA; MING |
OTTAWA
OTTAWA
OTTAWA
OTTAWA |
|
CA
CA
CA
CA |
|
|
Assignee: |
HUAWEI TECHNOLOGIES CO.,
LTD.
SHENZHEN
CN
|
Family ID: |
1000004793694 |
Appl. No.: |
16/854329 |
Filed: |
April 21, 2020 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
62939284 |
Nov 22, 2019 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H04L 41/0823 20130101;
H04W 92/10 20130101; H04L 41/16 20130101 |
International
Class: |
H04L 12/24 20060101
H04L012/24; H04W 92/10 20060101 H04W092/10 |
Claims
1. A method in a wireless communication network, the method
comprising: transmitting, by a first device, information regarding
an artificial intelligence or machine learning (AI/ML) capability
of the first device to a second device over an air interface
between the first device and the second device, the information
regarding an AI/ML capability of the first device identifying
whether the first device supports AI/ML for optimization of at
least one air interface component over the air interface.
2. The method of claim 1, wherein the information regarding an
AI/ML capability of the first device comprises information
indicating at least one of the following: the first device is
capable of supporting a type and/or level of complexity of AI/ML;
whether the first device assists with an AI/ML training process for
optimization of the at least one air interface component; at least
one component of the at least one air interface component for which
the first device supports AI/ML optimization.
3. The method of claim 2, wherein the at least one component of the
at least one air interface component includes at least one of a
coding component, a modulation component and a waveform
component.
4. The method of claim 2, wherein the information indicating at
least one component of the at least one air interface component for
which the first device supports AI/ML optimization further
comprises information indicating whether the first device supports
joint optimization of two or more air interface components.
5. The method of claim 1, wherein transmitting the information
regarding an AI/ML capability of the first device comprises at
least one of: transmitting the information in response to receiving
an enquiry; and transmitting the information as part of an initial
network access procedure.
6. A method in a wireless communication network, the method
comprising: receiving, by a second device, information regarding an
artificial intelligence or machine learning (AI/ML) capability of a
first device over an air interface between the first device and the
second device, the information regarding an AI/ML capability of the
first device identifying whether the first device supports AI/ML
for optimization of at least one air interface component over the
air interface; and transmitting an AI/ML training request to the
first device based at least in part on the information regarding
the AI/ML capability of the first device.
7. The method of claim 6, wherein the information regarding an
AI/ML capability of the first device comprises information
indicating at least one of the following: the first device is
capable of supporting a type and/or level of complexity of AI/ML;
whether the first device assists with an AI/ML training process for
optimization of the at least one air interface component; at least
one component of the at least one air interface component for which
the first device supports AI/ML optimization.
8. The method of claim 7, wherein the at least one component of the
at least one air interface component includes at least one of a
coding component, a modulation component and a waveform
component.
9. The method of claim 7, wherein the information indicating at
least one component of the at least one air interface component for
which the first device supports AI/ML optimization further
comprises information indicating whether the first device supports
joint optimization of two or more components of the at least one
air interface component.
10. The method of claim 6, wherein receiving the information
regarding an AI/ML capability of the first device comprises
receiving the information as part of an initial network access
procedure for the first device.
11. The method of claim 6, wherein transmitting the AI/ML training
request comprises transmitting the AI/ML training request through
downlink control information (DCI) on a downlink control channel or
RRC signaling or the combination of the DCI and RRC signaling.
12. The method of claim 11, further comprising, receiving a
training request response from the device confirming that the
device has transitioned to an AI/ML training mode.
13. The method of claim 6, further comprising: transmitting a
training termination signal to the first device to indicate that a
training phase has finished.
14. An apparatus comprising: at least one processor; and a computer
readable storage medium operatively coupled to the at least one
processor, the computer readable storage medium storing programming
for execution by the at least one processor, the programming
comprising instructions to: transmit, from the apparatus,
information regarding an artificial intelligence or machine
learning (AI/ML) capability of the apparatus to a network device
over an air interface between the appatus and the network device,
the information regarding an AI/ML capability of the apparatus
identifying whether the apparatus supports AI/ML for optimization
of at least one air interface component over the air interface.
15. The apparatus of claim 14, wherein the information regarding an
AI/ML capability of the apparatus comprises information indicating
at least one of the following: the apparatus is capable of
supporting a type and/or level of complexity of AI/ML; whether the
apparatus assists with an AI/ML training process for optimization
of the at least one air interface component; at least one component
of the at least one air interface component for which the apparatus
supports AI/ML optimization.
16. The apparatus of claim 15, wherein the at least one component
of the at least one air interface component includes at least one
of a coding component, a modulation component and a waveform
component.
17. The apparatus of claim 15, wherein the information indicating
at least one component of the at least one air interface component
for which the apparatus supports AI/ML optimization further
comprises information indicating whether the apparatus supports
joint optimization of two or more components of the at least one
air interface component.
18. A network apparatus comprising: at least one processor; and a
computer readable storage medium operatively coupled to the at
least processor, the computer readable storage medium storing
programming for execution by the at least processor, the
programming comprising instructions to: receive, by the network
apparatus information regarding an artificial intelligence or
machine learning (AI/ML) capability of a first device over an air
interface between the first device and the network apparatus, the
information regarding an AI/ML capability of the first device
identifying whether the first device supports AI/ML for
optimization of at least one air interface component over the air
interface; and transmit an AI/ML training request to the first
device based at least in part on the information regarding the
AI/ML capability of the first device.
19. The network apparatus of claim 18, wherein the information
regarding an AI/ML capability of the first device comprises
information indicating at least one of the following: the first
device is capable of supporting a type and/or level of complexity
of AI/ML; whether the first device assists with an AI/ML training
process for optimization of the at least one air interface
component; at least one component of the at least one air interface
component for which the first device supports AI/ML
optimization.
20. The network apparatus of claim 19, wherein the at least one
component of the at least one air interface component includes at
least one of a coding component, a modulation component and a
waveform component.
21. The network apparatus of claim 19, wherein the information
indicating at least one component of the at least one air interface
component for which the first device supports AI/ML optimization
further comprises information indicating whether the first device
supports joint optimization of two or more components of the at
least one air interface component.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] This application claims the benefit of U.S. Provisional
Patent Application No. 62/939,284 entitled "PERSONALIZED TAILORED
AIR INTERFACE" filed Nov. 22, 2019, the entire contents of which is
incorporated herein by reference.
FIELD
[0002] The present disclosure relates to wireless communication
generally, and, in particular embodiments, to methods and
apparatuses for air interface customization.
BACKGROUND
[0003] An air interface is the wireless communications link between
two or more communicating devices, such as an evolved NodeB (also
commonly referred to as a NodeB, a base station, NR base station, a
transmit point, a remote radio head, a communications controller, a
controller, and the like) and a user equipment (UE) (also commonly
referred to as a mobile station, a subscriber, a user, a terminal,
a phone, and the like). Typically, both communicating devices need
to know the air interface in order to successfully transmit and
receive a transmission.
[0004] In many wireless communication systems, the air interface
definition is a one-size-fits-all concept. The components within
the air interface cannot be changed or adapted once the air
interface is defined. In some implementations, only limited
parameters or modes of an air interface, such as a cyclic prefix
(CP) length or multiple input multiple output (MIMO) mode, can be
configured. In some modern wireless systems, a configurable air
interface concept has been adopted to provide a framework for a
more flexible air interface. It is intended to provide adaptation
of different components within the air interface, and to address
the potential requirements of future applications. Some modern
wireless systems, such as fifth generation (5G) or new radio (NR)
network systems, support network slicing, which is a network
architecture that enables the multiplexing of virtualized and
independent logical networks on the same physical network
infrastructure. In such systems, each network slice is an isolated
end-to-end network tailored to fulfill diverse requirements
requested by a particular service or application. A configurable
air interface has been proposed for NR networks that allows for
service or slice based optimization of the air interface to allow
the air interface to be configured based on a service or
application that will be supported by the air interface or the
network slice over which the service or application will be
provided.
SUMMARY
[0005] Different pairs of communicating devices (i.e., a
transmission sending device and a transmission receiving device)
may have different transmission capabilities and/or transmission
requirements. The different transmission capabilities and/or
transmission requirements typically cannot be met optimally by a
single air interface or air interface configuration.
[0006] The configurable air interface proposed for NR networks
allows service or slice based optimization based on selecting from
a predetermined subset of parameters or technologies for a
predetermined subset of air interface components. If the service
and/or network slice over which the service is provided changes,
the configurations of the components of the transmit and receive
chains of the communicating devices may be changed to match a new
predetermined service or slice specific air interface corresponding
to the new service or network slice.
[0007] However for each service, the transmission condition,
capability and requirements can still be quite different for each
device, which means, for example, that an air interface
configuration that may be optimal for delivering a service to one
device, for an example one UE, may not necessarily be optimal for
delivering the same service to another UE.
[0008] The present disclosure provides methods and apparatuses that
may be used to implement new air interfaces for wireless
communication that are tailored or personalized on a
device-specific basis, for example using artificial intelligence
and/or machine learning to provide device-specific air interface
optimization. For example, embodiments of the present disclosure
include new air interfaces that go beyond a network slice/service
specific air interface to a personalized tailored air interface
that includes a personalized service type and a personalized air
interface setting. Thus, using artificial intelligence and/or
machine learning to optimize a device-specific air interface, can
achieve a new air interface configuration to satisfy the
requirement of each UE on an individual basis.
[0009] One broad aspect of the present disclosure provides a method
in a wireless communication network in which a first device
transmits information regarding an artificial intelligence or
machine learning (AI/ML) capability of the first device to a second
device over an air interface between the first device and the
second device. For example, the information regarding an AI/ML
capability of the first device may identify whether the first
device supports AI/ML for optimization of at least one air
interface component over the air interface. Thus, the exchange of
AI/ML capability between two communicating devices is used to
optimize one or more air interface components to accomplish
device-specific air interface optimization.
[0010] Another broad aspect of the present disclosure provides a
method in a wireless communication network in which a second device
receives information regarding an artificial intelligence or
machine learning (AI/ML) capability of a first device over an air
interface between the first device and the second device. For
example, the information regarding an AI/ML capability of the first
device may identify whether the first device supports AI/ML for
optimization of at least one air interface component over the air
interface. In some embodiments, the second device may transmit an
AI/ML training request to the first device based at least in part
on the information regarding the AI/ML capability of the first
device. Thus, the exchange of AI/ML capability between two
communicating devices is used to optimize one or more air interface
components to accomplish device-specific air interface
optimization.
[0011] Yet another broad aspect of the present disclosure provides
an apparatus that includes at least one processor and a computer
readable storage medium operatively coupled to the at least one
processor. The computer readable storage medium stores programming
for execution by the at least one processor. The programming
includes instructions to transmit, from the apparatus, information
regarding an artificial intelligence or machine learning (AI/ML)
capability of the apparatus to a network device over an air
interface between the apparatus and the network device. For
example, the information regarding an AI/ML capability of the
apparatus may identify whether the apparatus supports AI/ML for
optimization of at least one air interface component over the air
interface.
[0012] Still another broad aspect of the present disclosure
provides a network apparatus that includes at least one processor
and a computer readable storage medium operatively coupled to the
at least one processor. The computer readable storage medium stores
programming for execution by the at least one processor. The
programming includes instructions to receive, by the network
apparatus, information regarding an artificial intelligence or
machine learning (AI/ML) capability of a first device over an air
interface between the first device and the network apparatus. For
example, the information regarding an AI/ML capability of the first
device may identify whether the first device supports AI/ML for
optimization of at least one air interface component over the air
interface. In some embodiment, the programming further comprises
instructions to transmit an AI/ML training request to the first
device based at least in part on the information regarding the
AI/ML capability of the first device.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Reference will now be made, by way of example, to the
accompanying drawings which show example embodiments of the present
application, and in which:
[0014] FIG. 1 is a schematic diagram of an example communication
system suitable for implementing examples described herein;
[0015] FIGS. 2 and 3 are block diagrams illustrating example
devices that may implement the methods and teachings according to
this disclosure;
[0016] FIG. 4 is a block diagram of an example computing system
that may implement the methods and teachings according to this
disclosure;
[0017] FIG. 5 illustrates an example air interface and components
thereof;
[0018] FIG. 6 is a block diagram of a transmit chain of a network
device and a receive chain of a user equipment device that are
configured to communicate over an air interface;
[0019] FIG. 7A is a block diagram of a first example of a transmit
chain of a network device and a receive chain of a user equipment
device that include machine learning components enabling
device-specific tailoring/customization of an air interface, in
accordance with a first embodiment of this disclosure;
[0020] FIG. 7B is a block diagram of a second example of a transmit
chain of a network device and a receive chain of a user equipment
device that include machine learning components enabling
device-specific tailoring/customization of an air interface, in
accordance with the first embodiment of this disclosure;
[0021] FIG. 8A is a block diagram of a first example of a transmit
chain of a network device and a receive chain of a user equipment
device that include machine learning components enabling
device-specific tailoring/customization of an air interface, in
accordance with a second embodiment of this disclosure;
[0022] FIG. 8B is a block diagram of a second example of a transmit
chain of a network device and a receive chain of a user equipment
device that include machine learning components enabling
device-specific tailoring/customization of an air interface, in
accordance with the second embodiment of this disclosure;
[0023] FIG. 9 is a block diagram of an example of a transmit chain
of a network device and a receive chain of a user equipment device
that include machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with a
third embodiment of this disclosure;
[0024] FIG. 10 is a block diagram of an example of a transmit chain
of a network device and a receive chain of a user equipment device
that include machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with a
fourth embodiment of this disclosure;
[0025] FIG. 11 is a block diagram of an example of a transmit chain
of a network device and a receive chain of a user equipment device
that include machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with a
fifth embodiment of this disclosure;
[0026] FIG. 12 illustrates an example of an over the air
information exchange procedure for a training phase of machine
learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure;
[0027] FIG. 13 illustrates another example of an over the air
information exchange procedure for a training phase of machine
learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure;
[0028] FIG. 14 illustrates an example of an over the air
information exchange procedure for a normal operations phase of
machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure; and
[0029] FIG. 15 illustrates an example of an over the air
information exchange procedure for a re-training phase of machine
learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure.
[0030] Similar reference numerals may have been used in different
figures to denote similar components.
DESCRIPTION OF EXAMPLE EMBODIMENTS
[0031] To assist in understanding the present disclosure, an
example wireless communication system is described below.
[0032] FIG. 1 illustrates an example wireless communication system
100 (also referred to as wireless system 100) in which embodiments
of the present disclosure could be implemented. In general, the
wireless system 100 enables multiple wireless or wired elements to
communicate data and other content. The wireless system 100 may
enable content (e.g., voice, data, video, text, etc.) to be
communicated (e.g., via broadcast, narrowcast, user device to user
device, etc.) among entities of the system 100. The wireless system
100 may operate by sharing resources such as bandwidth. The
wireless system 100 may be suitable for wireless communications
using 5G technology and/or later generation wireless technology
(e.g., 6G or later). In some examples, the wireless system 100 may
also accommodate some legacy wireless technology (e.g., 3G or 4G
wireless technology).
[0033] In the example shown, the wireless system 100 includes
electronic devices (ED) 110a-110c (generically referred to as ED
110), radio access networks (RANs) 120a-120b (generically referred
to as RAN 120), a core network 130, a public switched telephone
network (PSTN) 140, the internet 150, and other networks 160. In
some examples, one or more of the networks may be omitted or
replaced by a different type of network. Other networks may be
included in the wireless system 100. Although certain numbers of
these components or elements are shown in FIG. 1, any reasonable
number of these components or elements may be included in the
wireless system 100.
[0034] The EDs 110 are configured to operate, communicate, or both,
in the wireless system 100. For example, the EDs 110 may be
configured to transmit, receive, or both via wireless or wired
communication channels. Each ED 110 represents any suitable end
user device for wireless operation and may include such devices (or
may be referred to) as a user equipment/device (UE), a wireless
transmit/receive unit (WTRU), a mobile station, a fixed or mobile
subscriber unit, a cellular telephone, a station (STA), a machine
type communication (MTC) device, a personal digital assistant
(PDA), a smartphone, a laptop, a computer, a tablet, a wireless
sensor, or a consumer electronics device, among other
possibilities. Future generation EDs 110 may be referred to using
other terms.
[0035] In FIG. 1, the RANs 120 include base stations (BSs)
170a-170b (generically referred to as BS 170), respectively. Each
BS 170 is configured to wirelessly interface with one or more of
the EDs 110 to enable access to any other BS 170, the core network
130, the PSTN 140, the internet 150, and/or the other networks 160.
For example, the BS 170s may include (or be) one or more of several
well-known devices, such as a base transceiver station (BTS), a
radio base station, a Node-B (NodeB), an evolved NodeB (eNodeB), a
Home eNodeB, a gNodeB (sometimes called a next-generation Node B),
a transmission point (TP), a transmit and receive point (TRP), a
site controller, an access point (AP), or a wireless router, among
other possibilities. Future generation BSs 170 may be referred to
using other terms. Any ED 110 may be alternatively or additionally
configured to interface, access, or communicate with any other BS
170, the internet 150, the core network 130, the PSTN 140, the
other networks 160, or any combination of the preceding. The
wireless system 100 may include RANs, such as RAN 120b, wherein the
corresponding BS 170b accesses the core network 130 via the
internet 150, as shown.
[0036] The EDs 110 and BSs 170 are examples of communication
equipment that can be configured to implement some or all of the
functionality and/or embodiments described herein. In the
embodiment shown in FIG. 1, the BS 170a forms part of the RAN 120a,
which may include other BSs, base station controller(s) (BSC),
radio network controller(s) (RNC), relay nodes, elements, and/or
devices. Any BS 170 may be a single element, as shown, or multiple
elements, distributed in the corresponding RAN, or otherwise. Also,
the BS 170b forms part of the RAN 120b, which may include other
BSs, elements, and/or devices. Each BS 170 transmits and/or
receives wireless signals within a particular geographic region or
area, sometimes referred to as a "cell" or "coverage area". A cell
may be further divided into cell sectors, and a BS 170 may, for
example, employ multiple transceivers to provide service to
multiple sectors. In some embodiments there may be established pico
or femto cells where the radio access technology supports such. A
macro cell may encompass one or more smaller cells. In some
embodiments, multiple transceivers could be used for each cell, for
example using multiple-input multiple-output (MIMO) technology. The
number of RANs 120 shown is exemplary only. Any number of RANs may
be contemplated when devising the wireless system 100.
[0037] The BSs 170 communicate with one or more of the EDs 110 over
one or more air interfaces 190a using wireless communication links
(e.g. radio frequency (RF), microwave, infrared (IR), etc.). The
EDs 110 may also communicate directly with one another via one or
more sidelink air interfaces 190b. The interfaces 190a and 190b may
be generally referred to as air interfaces 190. BS-ED
communications over interfaces 190a and ED-ED communications over
interfaces 190b may use similar communication technology. The air
interfaces 190 may utilize any suitable radio access technology.
For example, the wireless system 100 may implement one or more
channel access methods, such as code division multiple access
(CDMA), time division multiple access (TDMA), frequency division
multiple access (FDMA), orthogonal FDMA (OFDMA), or single-carrier
FDMA (SC-FDMA) in the air interfaces 190. The air interfaces 190
may utilize other higher dimension signal spaces, which may involve
a combine of orthogonal and/or non-orthogonal dimensions.
[0038] The RANs 120 are in communication with the core network 130
to provide the EDs 110 with various services such as voice, data,
and other services. The RANs 120 and/or the core network 130 may be
in direct or indirect communication with one or more other RANs
(not shown), which may or may not be directly served by core
network 130, and may or may not employ the same radio access
technology as RAN 120a, RAN 120b or both. The core network 130 may
also serve as a gateway access between (i) the RANs 120 or EDs 110
or both, and (ii) other networks (such as the PSTN 140, the
internet 150, and the other networks 160). In addition, some or all
of the EDs 110 may include functionality for communicating with
different wireless networks over different wireless links using
different wireless technologies and/or protocols. Instead of
wireless communication (or in addition thereto), the EDs 110 may
communicate via wired communication channels to a service provider
or switch (not shown), and to the internet 150. PSTN 140 may
include circuit switched telephone networks for providing plain old
telephone service (POTS). Internet 150 may include a network of
computers and subnets (intranets) or both, and incorporate
protocols, such as Internet Protocol (IP), Transmission Control
Protocol (TCP), User Datagram Protocol (UDP). EDs 110 may be
multimode devices capable of operation according to multiple radio
access technologies, and incorporate multiple transceivers
necessary to support such.
[0039] FIGS. 2 and 3 illustrate example devices that may implement
the methods and teachings according to this disclosure. In
particular, FIG. 2 illustrates an example ED 110, and FIG. 3
illustrates an example base station 170. These components could be
used in the communication system 100 or in any other suitable
system.
[0040] As shown in FIG. 2, the ED 110 includes at least one
processing unit 200. The processing unit 200 implements various
processing operations of the ED 110. For example, the processing
unit 200 could perform signal coding, data processing, power
control, input/output processing, or any other functionality
enabling the ED 110 to operate in the communication system 100. The
processing unit 200 may also be configured to implement some or all
of the functionality and/or embodiments described in more detail
elsewhere herein. Each processing unit 200 includes any suitable
processing or computing device configured to perform one or more
operations. Each processing unit 200 could, for example, include a
microprocessor, microcontroller, digital signal processor, field
programmable gate array, or application specific integrated
circuit.
[0041] The ED 110 also includes at least one transceiver 202. The
transceiver 202 is configured to modulate data or other content for
transmission by at least one antenna or Network Interface
Controller (NIC) 204. The transceiver 202 is also configured to
demodulate data or other content received by the at least one
antenna 204. Each transceiver 202 includes any suitable structure
for generating signals for wireless or wired transmission and/or
processing signals received wirelessly or by wire. Each antenna 204
includes any suitable structure for transmitting and/or receiving
wireless or wired signals. One or multiple transceivers 202 could
be used in the ED 110. One or multiple antennas 204 could be used
in the ED 110. Although shown as a single functional unit, a
transceiver 202 could also be implemented using at least one
transmitter and at least one separate receiver.
[0042] The ED 110 further includes one or more input/output devices
206 or interfaces (such as a wired interface to the internet 150 in
FIG. 1). The input/output devices 206 permit interaction with a
user or other devices in the network. Each input/output device 206
includes any suitable structure for providing information to or
receiving information from a user, such as a speaker, microphone,
keypad, keyboard, display, or touch screen, including network
interface communications.
[0043] In addition, the ED 110 includes at least one memory 208.
The memory 208 stores instructions and data used, generated, or
collected by the ED 110. For example, the memory 208 could store
software instructions or modules configured to implement some or
all of the functionality and/or embodiments described herein and
that are executed by the processing unit(s) 200. Each memory 208
includes any suitable volatile and/or non-volatile storage and
retrieval device(s). Any suitable type of memory may be used, such
as random access memory (RAM), read only memory (ROM), hard disk,
optical disc, subscriber identity module (SIM) card, memory stick,
secure digital (SD) memory card, and the like.
[0044] As shown in FIG. 3, the base station 170 includes at least
one processing unit 1350, at least one transmitter 252, at least
one receiver 254, one or more antennas 256, at least one memory
258, and one or more input/output devices or interfaces 266. A
transceiver, not shown, may be used instead of the transmitter 252
and receiver 254. A scheduler 253 may be coupled to the processing
unit 250. The scheduler 253 may be included within or operated
separately from the base station 170. The processing unit 250
implements various processing operations of the base station 170,
such as signal coding, data processing, power control, input/output
processing, or any other functionality. The processing unit 250 can
also be configured to implement some or all of the functionality
and/or embodiments described in more detail herein. Each processing
unit 250 includes any suitable processing or computing device
configured to perform one or more operations. Each processing unit
250 could, for example, include a microprocessor, microcontroller,
digital signal processor, field programmable gate array, or
application specific integrated circuit.
[0045] Each transmitter 252 includes any suitable structure for
generating signals for wireless or wired transmission to one or
more EDs or other devices. Each receiver 254 includes any suitable
structure for processing signals received wirelessly or by wire
from one or more EDs or other devices. Although shown as separate
components, at least one transmitter 252 and at least one receiver
254 could be combined into a transceiver. Each antenna 256 includes
any suitable structure for transmitting and/or receiving wireless
or wired signals. Although a common antenna 256 is shown here as
being coupled to both the transmitter 252 and the receiver 254, one
or more antennas 256 could be coupled to the transmitter(s) 252,
and one or more separate antennas 256 could be coupled to the
receiver(s) 254. Each memory 258 includes any suitable volatile
and/or non-volatile storage and retrieval device(s) such as those
described above in connection to the ED 110 in FIG. 2. The memory
258 stores instructions and data used, generated, or collected by
the base station 170. For example, the memory 258 could store
software instructions or modules configured to implement some or
all of the functionality and/or embodiments described herein and
that are executed by the processing unit(s) 250.
[0046] Each input/output device 266 permits interaction with a user
or other devices in the network. Each input/output device 266
includes any suitable structure for providing information to or
receiving/providing information from a user, including network
interface communications.
[0047] It should be appreciated that one or more steps of the
embodiment methods provided herein may be performed by
corresponding units or modules, according to FIG. 4. For example, a
signal may be transmitted by a transmitting unit or a transmitting
module. A signal may be received by a receiving unit or a receiving
module. A signal may be processed by a processing unit or a
processing module. Other steps may be performed by an artificial
intelligence (AI) or machine learning (ML) module. The respective
units/modules may be implemented using hardware, one or more
components or devices that execute software, or a combination
thereof. For instance, one or more of the units/modules may be an
integrated circuit, such as field programmable gate arrays (FPGAs)
or application-specific integrated circuits (ASICs). It will be
appreciated that where the modules are implemented using software
for execution by a processor for example, they may be retrieved by
a processor, in whole or part as needed, individually or together
for processing, in single or multiple instances, and that the
modules themselves may include instructions for further deployment
and instantiation.
[0048] Additional details regarding the EDs such as 110 and base
stations such as 170 are known to those of skill in the art. As
such, these details are omitted here.
[0049] Referring back to FIG. 1, different pairs of communicating
devices (i.e., a transmission sending device and a transmission
receiving device), such as ED 110a communicating with BS 170a or ED
110b communicating with BS 170a, may have different transmission
capabilities and/or transmission requirements. The different
transmission capabilities and/or transmission requirements
typically cannot be met optimally by a single air interface or air
interface configuration.
[0050] As discussed above, a configurable air interface has been
proposed to address this issue. FIG. 5 illustrates a diagram of an
example of a configurable air interface 300. Air interface 300
comprises a number of building blocks that collectively specify how
a transmission is to be made and/or received. The building blocks
of air interface 300 may include waveform building block 305, frame
structure building block 310, multiple access scheme building block
315, a protocols building block 320, a coding and modulation
building block 325, and an antenna array processing building block
330.
[0051] Frame structure building block 310 may specify a
configuration of a frame or group of frames. Non-limiting examples
of frame structure options include a configurable multi-level
transmission time interval (TTI), a fixed TTI, a configurable
single-level TTI, a co-existence configuration, or configurable
slot, mini slot, or configurable symbol duration block (SDB) and
the like. The lengths of a TTI, slot, mini slot or SDB may also be
specified. Frame structure building block 310 may also or instead
specify timing parameters for DL and/or UL transmission, such as a
transmission period for DL and/or UL, and/or a time switch gap
between DL and UL transmissions. The frame structure can be for
various duplexing schemes, such as time domain duplexing (TDD),
frequency division duplexing (FDD) and full duplex operation.
[0052] Multiple access scheme building block 315 may specify how
access to a channel is scheduled or configured for one or more
users. Non-limiting examples of multiple access technique options
include scheduled access, grant-free access, dedicated channel
resource (no sharing between multiple users), contention based
shared channel resource, non-contention based shared channel
resource, cognitive radio based access, and the like.
[0053] Protocols building block 320 may specify how a transmission
and/or a re-transmission are to be made. Non-limiting examples of
transmission and/or re-transmission mechanism options include those
that specify a scheduled data pipe size, a signaling mechanism for
transmission and/or re-transmission, a re-transmission mechanism,
and the like.
[0054] Coding and modulation building block 325 may specify how
information being transmitted may be encoded (decoded) and
modulated (demodulated) for transmission (reception) purposes.
Non-limiting examples of coding and/or modulation technique options
include low density parity check (LDPC) codes, polar codes, turbo
trellis codes, turbo product codes, fountain codes, rateless codes,
network codes, binary phase shift keying (BPSK), .pi./2-BPSK,
quadrature phase shift keying (QPSK), quadrature amplitude
modulation (QAM) such as 16QAM, 64QAM, 256QAM, hierarchical
modulation, low PAPR modulation, non-linear modulation non-QAM
based modulation, and the like.
[0055] Waveform building block 305 may specify a shape and form of
a signal being transmitted. Non-limiting examples of waveform
options include Orthogonal Frequency Division Multiplexing (OFDM)
based waveform such as filtered OFDM (f-OFDM), Wavelet Packet
Modulation (WPM), Faster Than Nyquist (FTN) Waveform, low Peak to
Average Ratio Waveform (low PAPR WF such as DFT spread OFDM
waveform), Filter Bank Multicarrier (FBMC) Waveform, Single Carrier
Frequency Division Multiple Access (SC-FDMA), and the like. For
OFDM-based waveforms, the waveform building block 305 may specify
the associated waveform parameters such as sub-carrier spacings and
cyclic prefix (CP) overhead.
[0056] Antenna array processing building block 330 may specify
parameters for antenna array signal processing for channel
acquisition and precoding/beamforming generation. In some
embodiments, the functionality of the waveform building block 305
and the antenna array processing building block 330 may be combined
as a multiple antenna waveform generator block.
[0057] Since the air interface 300 comprises a plurality of
building blocks, and each building block may have a plurality of
candidate technologies, it may be possible to configure a large
number of different air interface profiles, where each air
interface profile defines a respective air interface configuration
option.
[0058] For example, the configurable air interface proposed for new
radio (NR) networks allows service or slice based optimization,
which can be advantageous because the potential application
requirements for air interface technologies can be complex and
diverse. Similar to the air interface 300 shown in FIG. 3, the
configurable air interface proposed for 5G networks supports
adaptive waveform, adaptive protocols, adaptive frame structure,
adaptive coding and modulation family and adaptive multiple access
schemes. With such mechanisms, the air interface can potentially
accommodate a wide variety of user services, spectrum bands and
traffic levels.
[0059] FIG. 6 illustrates an example of components in a transmit
chain 400 of a base station 170 and components of a receive chain
450 of a UE 110 that may be configurable as part of a configurable
air interface to allow the base station 170 and the UE 110 to
communicate.
[0060] The components of the transmit chain 400 of the base station
170 include a source encoder 402, a channel encoder 404 and a
modulator 406. Source encoder 402, channel encoder 404 and
modulator 406 may each be implemented as a specific hardware block,
or may be implemented in part as software modules executing in a
processor, such as a microprocessor, a digital signal processor, a
custom application specific integrated circuit, or a custom
compiled logic array of a field programmable logic array.
[0061] The components of the receive chain 450 of the UE 110
include a demodulator 452 and a channel decoder 454. Demodulator
452 and channel decoder 454 may each be implemented as a specific
hardware block, or may be implemented in part as software modules
executing in a processor, such as a microprocessor, a digital
signal processor, a custom application specific integrated circuit,
or a custom compiled logic array of a field programmable logic
array.
[0062] In operation, source encoder 402 encodes uncompressed raw
data to generate compressed information bits, which are in turn
encoded by channel encoder to generate channel coded information
bits, which are then modulated by modulator 406 to generate
modulated signals. In this example, the modulation performed by
modulator 406 includes quadrature amplitude modulation (QAM)
mapping and waveform generation. The modulated signals generated by
modulator 406 are transmitted from base station 170 to UE 110 over
one or more wireless channels. A base station can have multiple
transmit antennas, in which case a waveform may be generated for
each of the antennas. In such cases, the generated waveforms may
contain different contents for each of the multiple transmit
antennas, e.g., in a MIMO mode transmission. At UE 110, the
received signals from base station 170 are demodulated by
demodulator 452 to generate demodulated signals. A UE can have
multiple receive antennas, in which case demodulator 452 may be
configured to process waveforms received from multiple receive
antennas as part of the waveform recovery process. The demodulated
signals generated by demodulator 452 are decoded by channel decoder
454 to generate recovered compressed information bits. Source
decoder 456 decodes the recovered compressed information bits to
generate recovered uncompressed raw data.
[0063] Waveform here in the embodiment of FIG. 4 or the following
embodiments, may specify a shape and form of a signal being
transmitted. Non-limiting examples of waveform options include
Orthogonal Frequency Division Multiplexing (OFDM) based waveform
such as filtered OFDM (f-OFDM), Wavelet Packet Modulation (WPM),
Faster Than Nyquist (FTN) Waveform, low Peak to Average Ratio
Waveform (low PAPR WF such as DFT spread OFDM waveform), Filter
Bank Multicarrier (FBMC) Waveform, Single Carrier Frequency
Division Multiple Access (SC-FDMA), and the like. For OFDM-based
waveforms, the waveform may specify the associated waveform
parameters such as sub-carrier spacings and cyclic prefix (CP)
overhead.
[0064] The coding and modulation performed by the components of the
transmit chain 400 and the corresponding demodulation and decoding
performed by the components of the receive chain 450 may be
configured according to a modulation and coding scheme (MCS)
corresponding to a service or slice specific air interface in order
to support delivery of a service or application to UE 110 according
to the selected code scheme and modulation scheme. If the service
and/or network slice over which the service is provided changes,
the configurations of the components of the transmit and receive
chains of the base station 170 and UE 110 may be changed to match a
new predetermined service or slice specific air interface
corresponding to the new service or network slice. As noted above,
a service or slice specific air interface such as this, which is
based on selecting from a predetermined subset of parameters or
technologies for a predetermined subset of air interface
components, can potentially accommodate a wide variety of user
services, spectrum bands and traffic levels.
[0065] However for each service, the transmission condition and
requirements can still be quite different for each UE/device, which
means, for example, that an air interface configuration that may be
optimal for delivering a service to one UE/device may not
necessarily be optimal for delivering the same service to another
UE. Therefore, it would be desirable to provide further
optimization of a UE/device specific air interface
configuration.
[0066] Machine learning (ML) and artificial intelligence (AI)
approaches have been used for solving many difficult and complex
problems. To assist in understanding the present disclosure, some
background discussion of ML and AI is now provided. AI is an
emerging and fast-growing field thanks to the advances made in the
field of computer architecture and in particular general purpose
graphics processing units (GP-GPUs). A neural network, which is a
form of ML, may be considered as a type of fitting function. Deep
learning is one realization of a neural network, which contains
more than one interconnected layer of artificial neurons. To train
a deep neural network to fit a function (e.g., training using a
great amount of input samples and output samples), the weight and
threshold of each neuron are updated iteratively, so as to minimize
an overall loss function or maximize an overall reward function.
The iteration may be achieved by a gradient-descent or ascent
back-propagation algorithm over training samples, which may require
that the deep neural network architecture and the loss or reward
function be mathematically differentiable.
[0067] Trainability typically requires: a function set (the neural
network architecture) that defines an exploration space boundary
within which a gradient-descent algorithm may traverse; and one or
more loss (or reward) function(s) being differentiable with respect
to each neuron's coefficient (for gradient-ascent or descent
training) on that neural network architecture.
[0068] A deep neural network is often used for performing feature
capture, and for performing prediction. Feature capture serves to
extract useful information from a number of complex data, and this
may be considered a form of dimension reduction. Prediction
involves interpolation or extrapolation, to generate new data
(generally referred to as predicted or estimated data) from sample
data. Both these tasks may assume that the input data possess an
intrinsic autoregression characteristic. For example, a pixel of an
image usually has some relationship with its neighboring pixels. A
convolutional neural network (CNN) may be developed to use this
relationship to reduce the dimension of the data.
[0069] The present disclosure describes examples that may be used
to implement new air interfaces for wireless communication that are
tailored or personalized on a device-specific basis using AI/ML to
provide device-specific air interface optimization. For example,
embodiments of the present disclosure include new air interfaces
that go beyond a network slice/service specific air interface to a
personalized tailored air interface that includes a personalized
service type and a personalized air interface setting. Examples of
such personalized air interface settings may include one or more of
the following: customized code scheme and modulation scheme;
customized transmission scheme such as MIMO beamforming (BF),
including channel acquisition/reconstruction and precoding;
customized waveform type and associated parameters such as
customized pulse shapes and parameters such as roll-off factors of
an RRC pulse; customized frame structure; customized
transmission/retransmission scheme and associated parameters such
as product-code or inter-codebook or inter-TB 2D joint coding based
retransmission and parameters such as incremental parity bit size
and interleavers used; UE cooperation based retransmission and/or
customized transmit-receive point (TRP) layer/type.
[0070] In some embodiments, the personalized tailored air interface
parameters may be determined using AI/ML based on the physical
speed/velocity at which the device is moving, a link budget of the
device, the channel conditions of the device, one or more device
capabilities and/or a service type that is to be supported. In some
embodiments, the service type itself can be customized with
UE-specific service parameters, such as quality of service (QoS)
requirement(s), traffic pattern, etc.
[0071] In some embodiments, the personalized tailored air interface
parameters may be optimized on the fly with minimal signaling
overhead. For example, for 5G network implementations, the
parameters may be configured from predefined candidate parameter
sets. For next generation network implementations, e.g., for sixth
generation (6G) networks, the parameters maybe adapted in a more
flexible manner with real time or near real time optimization.
[0072] As will be discussed later, the level or type of air
interface optimization available to a device may depend on the
AI/ML capability of the device. If a user equipment has some AI/ML
capability, the UE can work together with network device(s) to
optimize its air interface (i.e., both sides of the air interface
apply AI/ML to optimize the air interface). A UE that has no AI/ML
capability may still help a network device to optimize an air
interface during a training phase and/or during a normal operation
phase by providing some types of measurement results to the network
device for use in training AI/ML component(s) of the network
device. For example, a high end AI/ML capable device may be able to
benefit from full scale self-optimization of each component of an
air interface (e.g., optimization of coding, modulation and
waveform generation, MIMO operation optimization). A lower end
AI/ML capable device may only be able to benefit from partial
self-optimization of less than all components of an air interface.
In some cases, a device may be dependent on centralized
learning/training (e.g., all learning is done centrally in the
network, such as at a base station). In other cases,
learning/training may be based on federated learning, which is a
machine learning technique that trains an algorithm across multiple
decentralized edge devices or servers holding local data samples,
without exchanging their data samples. In still other cases,
learning/training may also or instead involve device cooperative
learning.
[0073] As discussed above, an air interface generally includes a
number of components and associated parameters that collectively
specify how a transmission is to be made and/or received over a
wireless communications link between two or more communicating
devices. For example, an air interface may include one or more
components defining the waveform(s), frame structure(s), multiple
access scheme(s), protocol(s), coding scheme(s) and/or modulation
scheme(s) for conveying data over a wireless communications link.
The methods and devices disclosed herein provide a mechanism of
AI/ML enabled/assisted air interface personalized optimization that
supports different levels of per-UE/device based optimization. The
disclosed examples also provide over the air signaling mechanisms
to support per-UE/device based air interface function
optimization.
[0074] FIG. 7A illustrates a first example of a transmit chain 500
of a base station 170 and a receive chain 550 of a UE 110 that each
include an AI/ML module 502,552 that is trainable in order to
provide a tailored personalized air interface between the base
station 170 and UE 110, in accordance with an embodiment of the
present disclosure. AI/ML components as referenced herein are
intended to be modules or blocks based on an implementation of ML
mechanisms. One example of an ML implementation is a neural network
implemented in hardware, one or more components that execute
software, or a combination thereof.
[0075] The AI/ML module 502 of the base station 170 includes a
joint source and channel encoder component 504, a modulator
component 506 and a waveform generator component 508.
[0076] The AI/ML module 552 of the UE 110 includes a joint waveform
recovery, demodulator and source and channel decoder component
554.
[0077] The AI/ML module 502 provides AI/ML based autonomous
optimization of all basic baseband signal processing functions
including channel coding (or source coding plus channel coding) via
encoding component 504, modulation via modulation component 506 and
waveform generation via waveform generator 508. The base station
170 may have multiple transmit antennas, and in such embodiments
the waveform generator 508 may be configured to generate a waveform
for each of the transmit antennas. The AI/ML module 552 at the UE
110 provides the reciprocal based band processing functionality in
order to recover information bits/raw data from signals received
from the base station 170. The UE 110 may have multiple receive
antennas, and in such embodiments the AI/ML module 552 may be
configured to process waveforms received from multiple receive
antennas as part of the waveform recovery process.
[0078] The coding, modulation and waveform generation may be
optimized individually or two or more may be jointly optimized.
[0079] Several options are possible for individual optimization of
the various components of the AI/ML modules 502, 552. Some
non-limiting examples of these options are described below.
[0080] For example, for individual optimization of channel coding
without a predefined coding scheme and parameters,
self-learning/training and optimization may be used to determine an
optimal coding scheme and parameters. For example, in some
embodiments, a forward error correction (FEC) scheme is not
predefined and AI/ML is used to determine a UE specific customized
FEC scheme. In such embodiments, autoencoder based ML may be used
as part of an iterative training process during a training phase in
order to train an encoder component at a transmitting device and a
decoder component at a receiving device. For example, during such a
training process, an encoder at a base station and a decoder at a
UE may be iteratively trained by exchanging a training
sequence/updated training sequence. In general, the more trained
cases/scenarios, the better performance. After training is done,
the trained encoder component at the transmitting device and the
trained decoder component at the receiving device can work together
based on changing channel conditions to provide encoded data that
may outperform results generated from a non-AI/ML based FEC scheme.
In some embodiments, the AI/ML algorithms for
self-learning/training and optimization may be downloaded by the UE
from a network/server/other device.
[0081] For individual optimization of channel coding with
predefined coding schemes, such as low density parity check (LDPC)
code, Reed-Muller (RM) code, polar code or other coding scheme, the
parameters for the coding scheme can be optimized.
[0082] The parameters for channel coding can be signaled to UE from
time to time (periodically or event triggered), e.g., via radio
resource control (RRC) signaling or dynamically through downlink
control information (DCI) in a dynamic downlink control channel or
the combination of the RRC signaling and DCI, or group DCI, or
other new physical layer signaling. Training can be done all on the
network side or assisted by UE side training or mutual training
between the network side and the UE side.
[0083] In the example illustrated in FIG. 7A, the input to AI/ML
module 502 is uncompressed raw data and source coding and channel
coding are done jointly by AI/ML component 504. An alternative
example is illustrated in FIG. 7B, in which source coding is done
separately by a source encoder 501 to generate compressed
information bits that are then received by AI/ML module 502 where
they are channel coded by AI/ML component 504. Similarly, in the
example illustrated in FIG. 7B, the output of the AI/ML module 552
at the UE 110 is recovered compressed information bits that are
then decoded by a source decoder 555 to generate recovered raw
data, whereas the AI/ML module 552 in FIG. 7A outputs recovered raw
data.
[0084] For individual optimization of modulation without a
predefined constellation, modulation may be done by an AI/ML
module, the optimization targets and or algorithms of which (e.g.,
the AI/ML component 506) are understood by both the transmitter and
the receiver (e.g., the base station 170 and UE 110, respectively,
in the example scenario shown in FIG. 7A). For example, the AI/ML
algorithm may be configured to maximize Euclidian or non-Euclidian
distance between constellation points.
[0085] For individual optimization of modulation with a predefined
non-linear modulator, the parameters for the modulation may be done
by self-optimization.
[0086] For individual optimization of waveform generation without a
predefined waveform type, without a predefined pulse shape and
without predefined waveform parameters, self-learning/training and
optimization may be used to determine optimal waveform type, pulse
shape and waveform parameters. In some embodiments, the AI/ML
algorithms for self-learning/training and optimization may be
downloaded by the UE from a network/server/other device.
[0087] In some embodiments, there may be a finite set of predefined
waveform types, and selection of a predefined waveform type from
the finite set and determination of the pulse shape and other
waveform parameters may be done through self-optimization.
[0088] Several options are also possible for joint optimization of
two or more of the components of the AI/ML modules 502, 552. Some
non-limiting examples of these options are described below.
[0089] For example, in some embodiments, the coding via component
504 (channel coding or joint source and channel coding) and the
modulation implemented via component 506 may be jointly optimized
with AI/ML, and the waveform generation via component 508 may be
optimized separately. Multi-dimensional modulation, which is
conceptually similar to trellis-coded modulation, is one example of
a combined coding and modulation scheme that may be used in some
embodiments of the present disclosure. For example, in some
embodiments, AI/ML may be used to create a customized
multi-dimensional modulation scheme for a pair of communicating
devices, e.g., a base station and a UE.
[0090] In other embodiments, the modulation via component 504 and
the waveform generation via component 508 may be jointly optimized
with AI/ML, and the coding via component 504 may be optimized
separately. In other embodiments, the coding, modulation and
waveform generation may all be jointly optimized with AI/ML.
[0091] FIGS. 8A and 8B illustrate examples of a transmit chain 600
of a base station 170 and a receive chain 650 of a UE 110 that each
include an AI/ML module 602,652 that is trainable in order to
realize UE specific optimization and/or provide a tailored or
personalized air interface between the base station 170 and UE 110,
in accordance with a second embodiment of the present disclosure.
In the example shown in FIG. 8A, the transmit chain 600 of base
station 170 includes an AI/ML module 602 and a waveform generator
605. AI/ML module 602 of the base station 170 includes a joint
source and channel encoder and modulation component 604. Similarly,
in this example the receive chain 650 of UE 110 includes a waveform
processor 651 and an AI/ML module 652, which includes a joint
demodulator and source and channel decoder component 654.
[0092] Unlike the examples shown in FIGS. 7A and 7B, in which the
AI/ML modules 502,552 provide AI/ML based autonomous optimization
of all basic baseband signal processing functions including
coding/decoding, modulation/demodulation and waveform
generation/processing, in the example shown in FIG. 8A the AI/ML
module 602 provides AI/ML based autonomous optimization of coding
and modulation via component 604, and non-AI/ML based waveform
generation is managed independently via waveform generator 605. The
base station 170 may have multiple transmit antennas, and in such
embodiments the waveform generator 605 may be configured to
generate a waveform for each of the transmit antennas. The AI/ML
module 652 at the UE 110 provides the reciprocal optimized baseband
processing functionality on modulated signals recovered by waveform
processor 651. The UE 110 may have multiple receive antennas, and
in such embodiments the waveform processor 651 may be configured to
process waveforms received from multiple receive antennas as part
of the waveform recovery process.
[0093] In the example illustrated in FIG. 8A, the input to AI/ML
module 602 is uncompressed raw data and joint source and channel
coding and modulation are done by AI/ML component 604. The example
illustrated in FIG. 8B differs from the example illustrated in FIG.
8A in that in FIG. 8A source coding is done separately by a source
encoder 601 to generate information bits that are then received by
AI/ML module 602 where they are jointly channel coded and modulated
by AI/ML component 604. Similarly, in the example illustrated in
FIG. 8B, the output of the AI/ML module 652 at the UE 110 is
recovered compressed information bits that are then decoded by a
source decoder 655 to generate recovered raw data, whereas the
AI/ML module 652 in FIG. 8A outputs recovered raw data.
[0094] In the examples shown in FIGS. 8A and 8B, training of the
AI/ML modules 602 and 652 may be done by self-learning/training
optimization. Coding and modulation may be optimized by AI/ML
separately or jointly, as discussed earlier.
[0095] As mentioned above, in the examples shown in FIGS. 8A and
8B, waveform generation via waveform generator 605 at base station
170 and waveform processing via waveform processor 651 at UE 110,
may be managed without AI/ML. For example, waveform types and
waveform parameters may be predefined and a waveform may be
selected from a predefined set of candidate waveforms according to
transmission requirements, such as peak to average power ratio
(PAPR), frequency band, frequency localization, and the like.
Alternatively, the waveform type and waveform parameters may be
dynamically signaled to a UE via for example downlink control
information (DCI) or radio resource control (RRC) signaling. In
some embodiments, the predefined set of candidate waveforms may
include single-carrier waveform and multi-carrier waveforms.
Furthermore, the predefined set of candidate waveforms may include
multiple candidate waveforms that differ in terms of one or more
parameters. For example, there may be multiple candidate
single-carrier waveforms predefined, such as single carrier offset
QAM (OQAM) waveforms, with root-raised cosine pulse, and predefined
roll-off factors.
[0096] FIG. 9 illustrates an example of a transmit chain 700 of a
base station 170 and a receive chain 750 of a UE 110 that each
include an AI/ML module 702,752 that is trainable in order to
provide a tailored personalized air interface between the base
station 170 and UE 110, in accordance with a third embodiment of
the present disclosure.
[0097] In the example shown in FIG. 9, the transmit chain 700 of
base station 170 includes a source encoder 701, a channel encoder
703 and an AI/ML module 702 that includes a modulation component
704 and a waveform generator component 706. In this example the
receive chain 750 of UE 110 includes an AI/ML module 752, which
includes a waveform processor component 756 and a demodulator
component 754, a channel decoder 755 and a source decoder 757.
[0098] Unlike the previous examples shown in FIGS. 7A, 7B, 8A and
8B, the example shown in FIG. 9 utilizes non-AI/ML based source and
channel coding/decoding and AI/ML based modulation/demodulation and
waveform generation/processing. At UE 110, the waveform processor
component 756 and the demodulator component 754 of the AI/ML module
652 provide the reciprocal optimized modulated signal recovery and
demodulation functionality to recover modulated information bits.
The recovered modulated information bits are decoded by channel
decoder 755 to generate recovered compressed information bits,
which are in turn decoded by source decoder 757 to generate
recovered raw data.
[0099] In the example shown in FIG. 9, training of the AI/ML
modules 602 and 652 may be done by self-learning/training
optimization. Modulation and waveform generation may be optimized
by AI/ML separately or jointly, as discussed earlier. As mentioned
above, in the example shown in FIG. 9, source and channel coding
via source encoder 701 and channel encoder 703 at base station 170
and channel and source decoding via channel decoder 755 and source
decoder 757 at UE 110, may be managed without AI/ML. For example,
coding schemes and associated parameters may be predefined and a
coding scheme may be selected from a predefined set of coding
schemes according to a transmission requirement. Alternatively, the
coding scheme and associated parameters may be dynamically signaled
to a UE via for example downlink control information (DCI) or radio
resource control (RRC) signaling.
[0100] FIG. 10 illustrates an example of a transmit chain 800 of a
base station 170 and a receive chain 850 of a UE 110 that each
include an AI/ML module 802,852 that is trainable in order to
provide a tailored personalized air interface between the base
station 170 and UE 110, in accordance with a fourth embodiment of
the present disclosure.
[0101] In the example shown in FIG. 10, the transmit chain 800 of
base station 170 includes a source encoder 801, a channel encoder
803, an AI/ML module 802, which includes a modulation component
804, and a waveform generator 805. In this example the receive
chain 850 of UE 110 includes a waveform processor 851, an AI/ML
module 852, a channel decoder 855 and a source decoder 857. The
AI/ML module 852 includes a demodulator component 854.
[0102] Unlike the previous examples, the example shown in FIG. 10
utilizes non-AI/ML based channel coding/decoding and waveform
generation/processing and AI/ML based modulation/demodulation. At
UE 110, the waveform processor 851, channel decoder 855 and source
decoder 857 provide non AI/ML based signal recover, channel
decoding and source decoding, respectively, and the demodulator
component 854 of the AI/ML module 852 provides optimized
demodulation functionality that is the reciprocal of the modulation
functionality performed by the modulation component 804 at base
station 170.
[0103] For optimization of modulation without a predefined
constellation, an AI/ML algorithm implemented by modulation
component 804 may be configured to maximize Euclidian or
non-Euclidian distance between constellation points.
[0104] For optimization of modulation with a predefined non-linear
modulator, the parameters for the modulation may be done by
self-optimization, e.g., to optimize the distance of modulated
symbols. In some scenarios, non-AI/ML based optimization of
modulation may also or instead be utilized. As mentioned above, in
the example shown in FIG. 10, source and channel coding via source
encoder 801 and channel encoder 803 and waveform generation via
waveform generator 805 at base station 170 and waveform processing
via waveform processor 851 and channel and source decoding via
channel decoder 855 and source decoder 857 at UE 110, may be
managed without AI/ML. For example, waveform types and associated
parameters as well as coding schemes and associated parameters may
be predefined and a waveform type and a coding scheme may be
selected from predefined sets according to a transmission
requirement, as discussed previously. Alternatively, the coding
scheme and associated parameters and/or the waveform type and
waveform parameters may be dynamically signaled to a UE via for
example downlink control information (DCI) or radio resource
control (RRC) signaling.
[0105] FIG. 11 illustrates an example of a transmit chain 900 of a
base station 170 and a receive chain 950 of a UE 110 that each
include an AI/ML module 902,952 that is trainable in order to
provide a tailored personalized air interface between the base
station 170 and UE 110, in accordance with a fifth embodiment of
the present disclosure.
[0106] In the example shown in FIG. 11, the transmit chain 900 of
base station 170 includes a source encoder 901, a channel encoder
903, a QAM mapping component 905 and an AI/ML module 902 that
includes a waveform generation component 904. In this example the
receive chain 950 of UE 110 includes an AI/ML module 952, a QAM
demapping component 953, a channel decoder 955 and a source decoder
957. The AI/ML module 952 includes a waveform processing component
954.
[0107] Unlike the previous examples, the example shown in FIG. 11
utilizes non-AI/ML based source and channel coding/decoding and
modulation/demodulation and AI/ML based or assisted waveform
generation. The AI/ML based or assisted waveform generation may
enable per UE based optimization of one or more waveform
parameters, such as pulse shape, pulse width, subcarrier spacing
(SCS), cyclic prefix, pulse separation, sampling rate, PAPR and the
like.
[0108] For optimization of waveform generation without a predefined
waveform type, without a predefined pulse shape and without
predefined waveform parameters, self-learning/training and
optimization may be used to determine optimal waveform type, pulse
shape and waveform parameters. In some embodiments, the AI/ML
algorithms for self-learning/training and optimization may be
downloaded by the UE from a network/server/other device. In some
embodiments, there may be a finite set of predefined waveform
types, and selection of a predefined waveform type from the finite
set and determination of the pulse shape and other waveform
parameters may be done through self-optimization. In some
scenarios, non-AI/ML based optimization of waveform generation may
also or instead be utilized.
[0109] As mentioned above, in the example shown in FIG. 11, source
and channel coding via source encoder 901 and channel encoder 903
and modulation via QAM mapping component 905 at base station 170
and demodulation via QAM demapping component 953 and channel and
source decoding via channel decoder 955 and source decoder 957 at
UE 110, may be managed without AI/ML. For example, a modulation and
coding scheme and associated parameters may be selected from a
predefined set of modulation and coding schemes according to a
transmission requirement, as discussed previously. Alternatively,
the modulation and coding scheme and associated parameters may be
dynamically signaled to a UE via for example downlink control
information (DCI) or radio resource control (RRC) signaling.
[0110] Examples of over the air information exchange procedures
that may facilitate training of ML components of communicating
devices, such as various ML components of the base stations 170 and
UEs 110 of the examples shown in FIGS. 7 to 11 will now be
described with reference to FIGS. 12 to 14.
[0111] FIG. 12 is a signal flow diagram 1000 of an example of an
over the air information exchange procedure for a training phase of
machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure.
[0112] In the signal flow diagram 1000, a UE and a BS or other
network device are involved in an information exchange for an AI/ML
training phase 1150. Although only one UE and one BS are shown in
FIG. 12 to avoid congestion in the drawing, data collection or
information sharing during training, and similarly operation of a
communication network, are expected to involve more than one UE and
more than one BS. For example, in some embodiments training may be
done with the joint efforts from multiple network devices and
multiple UEs and air interface optimization may be done on a per UE
basis.
[0113] The information exchange procedure begins with UE sending
information indicating an AI/ML capability of the UE to the BS at
1010. The information indicating an AI/ML capability of the UE may
indicate whether or not the UE supports AI/ML for optimization of
an air interface. If the UE is capable of supporting AI/ML
optimization, the information may also or instead indicate what
type and/or level of complexity of AI/ML the UE is capable of
supporting, e.g., which function/operation AI/ML can be supported,
what kind of AI/ML algorithm can be supported (for example,
autoencoder, reinforcement learning, neural network (NN), deep
neural network (DNN), how many layers of NN can be supported,
etc.). In some embodiments, the information indicating an AI/ML
capability of the UE may also or instead include information
indicating whether the UE can assist with training.
[0114] In some embodiments, the information sent at 1010 may
include information indicating an AI/ML capability type of the UE.
The AI/ML capability type may identify whether the UE supports
AI/ML optimization of one or more components of the air interface
of the device. For example, the AI/ML capability type may be one of
a plurality of AI/ML capability types, where each AI/ML capability
type corresponds to support for a different level of AI/ML
capability. For example, the plurality of AI/ML capability types
may include an AI/ML capability type that indicates the UE supports
deep learning. As another example, the plurality of AI/ML
capability types may include different types that indicate
different combinations of air interface components that are
optimizable by AI/ML. For example, the plurality of AI/ML
capability types may include one or more of the following types:
[0115] a type corresponding to support for AI/ML based optimization
of all baseband signal processing components, such as coding
(channel coding or joint source and channel coding), modulation and
waveform generation (e.g., similar to the examples shown in FIGS.
7A and 7B); [0116] a type corresponding to support for AI/ML based
optimization of coding and modulation, but not waveform generation
(e.g., similar to the examples shown in FIGS. 8A and 8B); [0117] a
type corresponding to support for AI/ML based optimization of
modulation and waveform generation, but not coding (e.g., similar
to the example shown in FIG. 9); [0118] a type corresponding to
support for AI/ML based optimization of modulation, but not coding
and waveform generation (e.g., similar to the example shown in FIG.
10); [0119] a type corresponding to support for AI/ML based
optimization of waveform generation, but not coding and modulation
(e.g., similar to the example shown in FIG. 11).
[0120] In some embodiments, the information sent by the UE to the
BS at 1010 may be sent by the UE to the BS as part of an initial
access procedure to access the network. In other embodiments, the
information may also or instead be sent by the UE in response to a
capability enquiry from the BS (not shown).
[0121] After receiving AI/ML capability information from the UE
indicating that the UE supports AI/ML and can assist with training,
the BS sends a training request to the UE at 1012 to trigger a
training phase 1050. In some embodiments, the training request may
be sent to the UE through DCI (dynamic signaling) on a downlink
control channel or on a data channel. For example, in some
embodiments the training request may be sent to the UE as UE
specific or UE common DCI. For example, UE common DCI may be used
to send a training request to all UEs or a group of UEs.
[0122] The UE may send a response to the training request to the
BS, as indicated at 1014. This response may confirm that the UE has
entered a training mode. However, such a response can be optional
and may not be sent by a UE in some embodiments. At 1016 the BS
starts the training phase 1050 by sending a training signal that
includes a training sequence or training data to the UE. In some
embodiments, the BS may send a training sequence/training data to
the UE after a certain predefined time gap following transmission
of the training request at 1012. In other embodiments, the BS may
immediately transmit a training sequence/training data to the UE
after transmitting the training request at 1012. In still other
embodiments, the BS may wait until it has received a response to
the training request from the UE before transmitting the training
sequence/training data to the UE.
[0123] Non-limiting examples of channels that may be used by the BS
to send training sequences/training data to UE include: [0124]
Dynamic control channel: When the number of bits required to send
the training sequence/training data is less than a certain
threshold, a dynamic control channel may be used to send the
training sequence/training data. In some embodiment, several levels
of bit lengths may be defined. The different bit lengths may
correspond to different DCI formats or different DCI payloads. The
same DCI can be used for carrying training sequences/data for
different AI/ML modules. In some embodiments, a DCI field may
contain information indicating an AI/ML module the training
sequence/training data is to be used to train. [0125] Data channel:
In some embodiments, a data channel may be used to carry a training
sequence/training data. In such embodiments, the payload of the
data channel depends on the training sequence length or the amount
of training data that is to be sent. The DCI used to schedule such
a data channel can carry the information required for decoding the
data channel and AI/ML module indicator(s) to indicate which AI/ML
module(s) the training sequence/data is for. [0126] RRC channel: In
some embodiments, training sequences/training data can be sent to
UE via RRC signaling.
[0127] For its part, the UE starts to search for a training signal
(e.g., a training sequence or training data) sent by the network
after sending back a response to the training request at 1014 or
after receiving the training request at 1012 with or without a
predefined time gap. The channel resource and the transmission
parameters for the training signal, such as MCS and demodulation
reference signal (DMRS), can be predefined or preconfigured (for
example by RRC signaling) or signaled by dynamic control signaling
(similar to the detection of DCI for a scheduled data channel). In
some embodiments, the training sequence/training data may be
carried in a dynamic control channel directly (e.g., certain bits
in a dynamic control channel may be reserved for carrying training
sequence/training data).
[0128] At 1018 the UE sends a training response message to the BS
that includes feedback information based on processing of the
received training signal. In some embodiments, the training
response message may include feedback information indicating an
updated training sequence for an iterative training process (e.g.,
for autoencoder based ML) or certain type(s) of measurement results
to help Tx/Rx to further train or refine the training of a NN,
e.g., for enforcement learning. In some embodiments, such
measurements may include, for example, the error margin obtained by
the UE in receiving the training sequence/data from the BS. For
example the measurement results may include information indicating
the mean square of errors and/or an error direction (e.g., error
increase or decrease). In some embodiments, the training response
message may also or instead include other feedback information,
such as an adjustment step size and direction (e.g., increase or
decrease by X amount, where X is the adjustment step size). In some
cases, the measurement results or feedback may be provided
implicitly. For example the adjustment of beamforming can be
indicated by the beam direction of the feedback signal. In some
embodiments, the training response message may be sent by the UE
through an uplink (UL) control channel. In other embodiments, the
training response message may be partially or entirely sent through
an UL data channel.
[0129] An AI/ML module that includes one or more AI/ML components,
such as a neural network, is trained in the network based on the
received training response message from the UE. In FIG. 12, this
training is indicated at 1019. For example, the parameters of an
AI/ML module, such as neural network weights, may be
updated/modified based on measurement results returned by the UE.
In some embodiments this training may be performed at least in part
in the BS, while in other embodiments the training may be performed
in part or in whole by another network device, such as a
centralized AI/ML server (not shown). At 1020, the BS sends
information to the UE to update AI/ML parameters, such as neural
network weights, in order to optimize one or more aspects of the
air interface between the UE and BS. In some embodiments this
training process is done iteratively, as indicated at 1040, whereby
the BS repeatedly transmits training sequence/data and iteratively
refines AI/ML parameters based on training response messages from
the UE. In some embodiments this iterative process may continue
until one or more target criteria is satisfied or until a
predefined number of iterations have occurred. It should be noted
that not all embodiments involve AI/ML functionality at UEs and
therefore AI/ML parameters need not necessarily be signaled to a UE
in all embodiments. At 1022, the BS terminates the training process
by sending a termination signal to the UE indicating the training
phase is finished, in response to which the UE transitions to a
normal operation phase 1060. In some embodiments, the training
termination signal may be transmitted to the UE through dynamic
signaling. In the normal operations phase 1060 the UE and BS may
then communicate via the updated air interface.
[0130] In some embodiments, the information exchange procedure
shown in FIG. 12 occurs at least partially in the Radio Resource
Control (RRC) layer.
[0131] In some embodiments, the information exchange procedure
shown in FIG. 12 occurs at least partially in a Medium Access
Control (MAC) layer. For example, the information exchange
signaling may be carried by a MAC control element (MAC CE)
implemented as a special bit string in a logical channel ID (LCID)
field of a MAC header.
[0132] In the example embodiment shown in FIG. 12, the AI/ML
training is performed in the network and the results of the
training are sent to the UE, which may be referred to as network
oriented training. In other embodiments, training may take place
jointly at the UE and in the network.
[0133] FIG. 13 is a signal flow diagram 1100 of an example of an
over the air information exchange procedure for a training phase of
machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure in which the training takes place
jointly at the UE and BS.
[0134] In the signal flow diagram 1100, a UE and a BS or other
network device are involved in an information exchange for an AI/ML
training phase 1150. The information exchange procedure begins with
the UE sending information indicating an AI/ML capability of the UE
to the BS at 1110. The information indicating an AI/ML capability
of the UE may include the same or similar information to that
described above with reference to the example embodiment shown in
FIG. 12, but in this example the information further also indicates
that the UE is capable of joint AI/ML training with the
network.
[0135] In some embodiments, the information sent by the UE to the
BS at 1110 may be sent as part of an initial access procedure to
access the network. In other embodiments, the information may also
or instead be sent by the UE in response to a capability enquiry
from the BS (not shown).
[0136] After receiving AI/ML capability information from the UE
indicating that the UE supports network and UE joint AI/ML
training, the BS sends a training request to the UE at 1112 to
trigger a training phase 1150. In some embodiments, the training
request may be sent to the UE through DCI (dynamic signaling) on a
downlink control channel or on a data channel. For example, in some
embodiments the training request may be sent to the UE with UE
specific or UE common DCI. For example, UE common DCI may be used
to send a training request to all UEs or a group of UEs. In some
embodiments, the training request may be set to the UE via RRC
signaling. In some embodiments, the training request may include
initial training setting(s)/parameter(s), such as initial NN
weights.
[0137] In some embodiments, the BS may also send AI/ML related
information to the UE to facilitate joint training such as: [0138]
Information indicating which AI/ML module is to be trained if there
is more than one AI/ML module that is trainable; [0139] Information
about the AI/ML algorithm and initial setting/parameters.
[0140] The AI/ML related information may be sent as part of the
training request sent at 1112 or may be sent separately from the
training request. The AI/ML related information sent to the UE,
such as information indicating AI/ML algorithm(s) and
setting/parameters, may have been selected by the BS or another
network device based at least in part on the AI/ML capability
information received from the UE. In some embodiments, the AI/ML
related information may include an instruction for the UE to
download initial AI/ML algorithm(s) and/or setting(s)/parameter(s),
in response to which the UE may then download initial AI/ML
algorithms and/or setting(s)/parameter(s) in accordance with the
instruction.
[0141] In some embodiments, after the UE has received the training
request and initial training information from the network, the UE
may send a response to the training request to the BS, as indicated
at 1114 in FIG. 13. This response may confirm that the UE has
entered a training mode. However, such a response can be optional
and may not be sent by a UE in some embodiments.
[0142] At 1116 the BS starts the training phase 1150 by sending a
training signal that includes a training sequence or training data
to the UE. In some embodiments, the BS may send a training
sequence/training data to the UE after a certain predefined time
gap following transmission of the training request at 1112. In
other embodiments, the BS may immediately transmit a training
sequence/training data to the UE after transmitting the training
request at 1112. In still other embodiments, the BS may wait until
it has received a response to the training request from the UE
before transmitting the training sequence/training data to the UE.
As noted above, in some embodiments the BS notifies the UE which
AI/ML module(s)/component(s) is/are to be trained by including
information in the training request that identifies one or more
AI/ML modules/components or by sending such information to the UE
in a separate communication. By doing so, the BS informs the UE
which AI/ML modules(s)/component(s) is/are to be trained based on
the training signal transmitted by the BS at 1116. Non-limiting
examples of channels that may be used by the BS to send training
sequences or training data to UE include those discussed above with
reference to FIG. 12, namely a dynamic control channel, a data
channel and/or RRC channel.
[0143] Similar to the UE in the example embodiment show in FIG. 12,
the UE in the example embodiment shown in FIG. 13 may start to
search for a training signal (e.g., a training sequence or training
data) after sending back a response to the training request at 1114
or after receiving the training request at 1112 with or without a
predefined time gap. The channel resource and the transmission
parameters for the training signal, such as MCS and DMRS, can be
predefined or preconfigured (e.g., by RRC signaling) or signaled by
dynamic control signaling. In some embodiments, the training
sequence/training data may be carried in a dynamic control channel
directly (e.g., certain bits in a dynamic control channel may be
reserved for carrying training sequence/training data).
[0144] At 1118 the UE sends a training response message to the BS.
In some embodiments, the training response message may include
feedback information indicating an updated training sequence for an
iterative training process (e.g., for autoencoder based ML) or
certain type(s) of measurement results to help further train or
refine the training of a NN, e.g., for enforcement learning. In
some embodiments, such measurements may include, for example, the
error margin obtained by the UE in receiving the training
sequence/data from the BS. For example the measurement results may
include information indicating the mean square of errors and/or an
error direction (e.g., error increase or decrease). In some
embodiments, the training response message may also or instead
include other feedback information, such as an adjustment step size
and direction (e.g., increase or decrease by X amount, where X is
the adjustment step size). In some cases, the measurement results
or feedback may be provided implicitly. For example the adjustment
of beamforming can be indicated by the beam direction of the
feedback signal. In some embodiments, the training response message
may be sent by the UE through an uplink (UL) control channel. In
other embodiments, the training response message may be partially
or entirely sent through an UL data channel.
[0145] In this embodiment, training of an AI/ML module that
includes one or more AI/ML components takes place jointly in the
network and at the UE, as indicated at 1119 in FIG. 13. For
example, parameters of an AI/ML module, such as neural network
weights, may be updated/modified based on measurement results
returned by the UE for the training sequence/data that was
transmitted by the BS.
[0146] In some embodiments, the UE and BS exchange updates of the
training setup and parameters, such as neural network weights, in
order to optimize one or more aspects of the air interface between
the UE and BS, as indicated at 1120 in FIG. 13. In other
embodiments, the UE and/or the BS may be able to update the
training setup and parameters autonomously based on their own
training process at 1119 without the further information exchange
indicated at 1120.
[0147] In some embodiments this training process is done
iteratively, as indicated at 1140, whereby the BS repeatedly
transmits training sequence/data and the UE and BS iteratively
refine AI/ML parameters based on training response messages from
the UE. In some embodiments this iterative process may continue
until one or more target criteria is satisfied or until a
predefined number of iterations have occurred. In some embodiments,
the training sequence/data may be updated during the iterative
training process.
[0148] At 1122, the BS terminates the training process by sending a
termination signal to the UE indicating the training phase is
finished, in response to which the UE transitions to a normal
operation phase 1160. In some embodiments, the UE may initiate
termination of the training phase by sending a termination
recommendation signal to the BS. In the normal operations phase
1160 the UE and BS may then communicate via the updated air
interface.
[0149] In some embodiments, the AI/ML algorithms and/or parameters
may have been pre-downloaded by the UE. In some embodiments, the
AI/ML capability information the UE sends at 1110 may include
information indicating pre-downloaded AI/ML algorithms and
parameters. In such embodiments, the BS may transmit a download
instruction to a UE to instruct the UE to download a selected AI/ML
algorithm or parameters if the AI/ML capability information
received from the UE indicates the selected AI/ML algorithm or
parameters have not been pre-downloaded by the UE.
[0150] In some embodiments, the information exchange procedure
shown in FIG. 13 occurs at least partially in the RRC layer.
[0151] In some embodiments, the information exchange procedure
shown in FIG. 13 occurs at least partially in a MAC layer. For
example, the information exchange signaling may be carried by a MAC
CE implemented as a special bit string in a LCID field of a MAC
header.
[0152] It should be understood that the specific AI/ML component
architectures that may be used in embodiments of the present
disclosure may be designed based on the particular application. For
example, where an AI/ML component is implemented with a deep neural
network (DNN), the specific DNN architecture that should be used
for a given application (e.g., joint coding and modulation
optimization or individual waveform generation optimization) may be
standardized (e.g., in agreed upon industry standards). For
example, standardization may include a standard definition of the
type(s) of neural network to be used, and certain parameters of the
neural network (e.g., number of layers, number of neurons in each
layer, etc.). Standardization may be application-specific. For
example, a table may be used to list the standard-defined neural
network types and parameters to be used for specific applications.
In the context of the wireless system 100 of FIG. 1, standardized
definitions may be stored in the memory of the BS 170, to enable
the BS 170 to select the appropriate DNN architecture and
parameters to be trained for a particular wireless communication
scenario.
[0153] As discussed above with respect to FIGS. 12 and 13, training
of DNN(s) (e.g., a single DNN implementing coding, modulation
and/or waveform generation, or separate DNNs for each) or other
AI/ML components may be performed at a BS or jointly at a BS and a
UE, and may be performed at the time of initial setup and
association between the BS and UE. In some examples, it may be
sufficient for the BS and/or the UE to train an AI/ML component,
e.g., DNN(s), at the time of setup. As well, training or
re-training may also be performed on-the-fly, for example in
response to significant change in the UE or BS and/or the
environment (e.g., addition of new UE(s), disassociation of a UE,
significant change in UE mobility, change in UE state or
significant change in channel, among other possibilities).
[0154] In some examples, training of the AI/ML components, such as
DNNs, at the BS and/or UE may be performed offline, for example
using data collected by the BS or UE. The collected data may
represent different wireless communication scenarios, such as
different times of day, different days of the week, different
traffic levels, etc. Training may be performed for a particular
scenario, to generate different sets of DNN weights for different
scenarios. The different sets of weights may be stored in
association with the different specific scenarios (e.g., in a
look-up table), for example in the memory of the BS or UE. The BS
or UE may then select and use a particular set of weights for the
DNN(s), in accordance with the specific scenario. For example, the
BS or UE may determine that it is handling communications for a
weekend evening (e.g., using information from an internal clock
and/or calendar) and use the corresponding set of weights to
implement the DNN(s) for coding, modulation and/or waveform
generation. This would result in the transmitter of the BS 170
performing coding, modulation and/or waveform generation suitable
for a weekend evening.
[0155] In some embodiments, offline and on-the-fly training may be
applied jointly. For example, on-the-fly re-trainining may be
performed to update training that was previously performed offline.
For example, a BS and/or UE may also retrain AI/ML components such
as DNN(s) on-the-fly, in response to dynamic changes in the
environment and/or in the UE or BS, as discussed above. Thus, the
BS or UE may update the table of weights dynamically. In some
examples, the table of weights may include sets of weights that are
standardized (e.g., defined in standards for very common scenarios)
and may also include sets of weights that are generated offline
and/or on-the-fly for certain scenarios.
[0156] The BS may provide an indexed table of weights and
associated scenarios to the UE. The BS may instruct the UE a
selected set of weights to use, for example by indicating the
corresponding index of a selected set of weights. The BS and/or UE
may retrain their AI/ML components and update their tables of
weights (e.g., in response to a new scenario) and communicate the
updated tables to one another, e.g., on a periodic or aperiodic
basis.
[0157] FIG. 14 is a signal flow diagram 1200 of an example of an
over the air information exchange procedure for a normal operations
phase 1260 of machine learning components enabling on-the-fly
device-specific tailoring/customization of an air interface, in
accordance with an embodiment of this disclosure. In this
embodiment, the on-the-fly update of AI/ML parameters may be
triggered by the network during the normal operation phase 1260, as
indicated at 1210. The network may trigger the on-the-fly update by
sending updated AI/ML parameters, such as DNN weights. In this
embodiment the on-the-fly update may also or instead be triggered
by the UE during the normal operation phase 1260, as indicated at
1212. The UE may trigger the on-the-fly update by sending updated
AI/ML parameters, such as DNN weights to the BS if the UE is
capable of self-training. Otherwise, the trigger that the UE sends
the BS at 1212 may simply comprise a request for an update from the
BS. In addition or instead of being triggered by the BS and/or the
UE, an on-the-fly update during the normal operation phase 1260 may
occur on a periodic or aperiodic basis, and may involve a mutual
information update exchange, as indicated at 1214.
[0158] FIG. 15 is a signal flow diagram 1300 of an example of an
over the air information exchange procedure for a re-training phase
of machine learning components enabling device-specific
tailoring/customization of an air interface, in accordance with an
embodiment of this disclosure.
[0159] In the signal flow diagram 1300, a UE and a BS or other
network device are involved in an information exchange for an AI/ML
re-training phase 1350. In this embodiment, the re-training phase
may be triggered by the network, as indicated at 1310. In some
embodiments, the BS may trigger the re-training by sending a
training request to the UE, e.g., through DCI, RRC or MAC signaling
as discussed earlier with reference to FIGS. 12 and 13. In this
embodiment the re-training phase may also or instead be triggered
by the UE, as indicated at 1312. In either case, during the
re-training phase 1350 the UE and BS exchange re-training signaling
as indicated at 1314 in order to facilitate re-training of AI/ML
components in the network and/or at the UE. For example, in some
embodiments the re-training signaling may include information
exchanges and signaling such as that indicated at 1016, 1018 and
1020 in FIG. 12 or at 1116, 1118 and 1120 in FIG. 13. In some
embodiments, re-training of an AI/ML module that includes one or
more AI/ML components may take place in the network or jointly in
the network and at the UE, as indicated at 1319 in FIG. 15.
[0160] In some embodiments this re-training process is done
iteratively, as indicated at 1340, whereby the BS repeatedly
transmits training sequence/data and the UE and BS iteratively
refine AI/ML parameters based on re-training response messages from
the UE. In some embodiments this iterative process may continue
until one or more target criteria is satisfied or until a
predefined number of iterations have occurred. In some embodiments,
the re-training sequence/data may be updated during the iterative
re-training process.
[0161] At 1316, the BS terminates the re-training process by
sending a termination signal to the UE indicating the re-training
phase is finished, in response to which the UE transitions to a
normal operation phase 1360. In some embodiments, the UE may
instead initiate termination of the re-training phase by sending a
termination recommendation signal to the BS, as indicated at 1318.
In the normal operations phase 1360 the UE and BS may then
communicate via the updated air interface resulting from the
re-training.
[0162] The above discussion refers to examples where the network
side training is performed by the BS. In other examples, AI/ML
component training may not be performed by the BS. For example,
referring again to FIG. 1, training may be performed by the core
network 130 or elsewhere in the wireless system 100 (e.g., using
cloud computing). A BS 170 may simply collect the relevant data and
forward the data to the appropriate network entity (e.g., the core
network 130) to perform the necessary training. The trained AI/ML
component parameters, e.g., weights of trained DNN(s), may then be
provided to the BS 170 and ED(s) 110.
[0163] Although the above discussion is in the context of the BS
170 in the role of a transmitter and the ED 110 in the role of a
receiver, it should be understood that the transmitter and receiver
roles may be reversed (e.g., for uplink communications). Further,
it should be understood that the transmitter and receiver roles may
be at two or more EDs 110a, 110b, 110c (e.g., for sidelink
communications). The BS 170 (or core network 130 or other network
entity) may perform the DNN training and may provide the trained
weights to the ED 110 in order for the ED 110 to implement the
DNN(s) for communicating with the BS 170.
EXAMPLE EMBODIMENTS
[0164] The following provides a non-limiting list of additional
Example Embodiments of the present disclosure:
Example Embodiment 1. A method in a wireless communication network,
the method comprising:
[0165] transmitting, by a first device, information regarding an
artificial intelligence or machine learning (AI/ML) capability of
the first device to a second device over a single air interface
between the first device and the second device, the information
regarding an AI/ML capability of the first device identifying
whether the first device supports AI/ML for optimization of at
least one air interface configuration over the single air
interface.
Example Embodiment 2. The method of Example Embodiment 1, wherein
the information regarding an AI/ML capability of the first device
comprises information indicating the first device is capable of
supporting a type and/or level of complexity of AI/ML. Example
Embodiment 3. The method of Example Embodiment 1 or 2, wherein the
information regarding an AI/ML capability of the first device
comprises information indicating whether the first device assists
with an AI/ML training process for optimization of the at least one
air interface configuration. Example Embodiment 4. The method of
any of Example Embodiments 1 to 3, wherein the information
regarding an AI/ML capability of the first device comprises
information indicating at least one component of the at least one
air interface configuration for which the first device supports
AI/ML optimization. Example Embodiment 5. The method of Example
Embodiment 4, wherein the at least one component of the at least
one air interface configuration includes at least one of a coding
component, a modulation component and a waveform component. Example
Embodiment 6. The method of Example Embodiment 4 or 5, wherein the
information indicating at least one component of the at least one
air interface configuration for which the first device supports
AI/ML optimization further comprises information indicating whether
the first device supports joint optimization of two or more
components of the at least one air interface configuration. Example
Embodiment 7. The method of any of Example Embodiments 1 to 6,
wherein transmitting the information regarding an AI/ML capability
of the first device comprises at least one of:
[0166] transmitting the information in response to receiving an
enquiry; and
[0167] transmitting the information as part of an initial network
access procedure.
Example Embodiment 8. The method of any of Example Embodiments 1 to
7, further comprising:
[0168] receiving an AI/ML training request from the second device;
and
[0169] after receiving the AI/ML training request, transitioning to
an AI/ML training mode.
Example Embodiment 9. The method of Example Embodiment 8, wherein
receiving the AI/ML training request comprises receiving the AI/ML
training request through downlink control information (DCI) on a
downlink control channel or RRC signaling or the combination of the
DCI and RRC signaling. Example Embodiment 10. The method of Example
Embodiment 8 or 9, further comprising, transmitting a training
request response to the second device to confirm that the first
device has transitioned to the AI/ML training mode. Example
Embodiment 11. The method of any of Example Embodiments 1 to 10,
further comprising receiving a training signal from the second
device that includes a training sequence or training data for
training at least one AI/ML module responsible for one or more
components of the at least one air interface configuration. Example
Embodiment 12. The method of Example Embodiment 11, wherein
receiving the training signal comprises receiving the training
signal on a dynamic control channel. Example Embodiment 13. The
method of Example Embodiment 12, wherein the dynamic control
channel includes a dynamic control information (DCI) field
containing information indicating an AI/ML module that is to be
trained. Example Embodiment 14. The method of Example Embodiment
11, wherein receiving the training signal comprises receiving the
training signal on a scheduled data channel, the method further
comprising receiving scheduling information for the data channel on
a dynamic control channel that includes a DCI field containing
information indicating an AI/ML module that is to be trained.
Example Embodiment 15. The method of any of Example Embodiments 11
to 14, further comprising, after receiving the training signal,
transmitting a training response message to the second device, the
training response message including feedback information based on
processing of the received training signal at the first device.
Example Embodiment 16. The method of Example Embodiment 15, wherein
the feedback information included in the training response message
includes an updated training sequence for an iterative training
process. Example Embodiment 17. The method of Example Embodiment 15
or 16, wherein the feedback information included in the training
response message includes measurement results based on the received
training signal. Example Embodiment 18. The method of Example
Embodiment 17, wherein the measurement results include an error
margin obtained by the first device in receiving the training
signal from the second device. Example Embodiment 19. The method of
any of Example Embodiments 15 to 18, further comprising, after
transmitting the training response message, receiving AI/ML update
information from the second device, the AI/ML update information
including information indicating updated AI/ML parameters for an
AI/ML module based on the feedback information provided by the
first device. Example Embodiment 20. The method of Example
Embodiment 19, further comprising, updating the AI/ML module in
accordance with the updated AI/ML parameters in order to update the
at least one air interface configuration for receiving
transmissions from the second device. Example Embodiment 21. The
method of any of Example Embodiments 15 to 18, further
comprising:
[0170] training one or more AI/ML modules at the first device based
on the training signal received from the second device; and
[0171] transmitting AI/ML update information to the second device,
the AI/ML update information including information indicating
updated AI/ML parameters for at least one of the one or more AI/ML
modules based on the training performed by the first device.
Example Embodiment 22. The method of Example Embodiment 21, further
comprising receiving AI/ML update information from the second
device, the AI/ML update information from the second device
including information indicating updated AI/ML parameters for at
least one of the one or more AI/ML modules based on training of one
or more AI/ML modules at the second device based on feedback
information provided in the training response message. Example
Embodiment 23. The method of Example Embodiment 22, further
comprising, updating the at least one air interface configuration
for receiving transmissions from the second device by updating the
one or more AI/ML modules in accordance with the updated AI/ML
parameters based on the training performed by the first device and
the updated AI/ML parameters received from the second device.
Example Embodiment 24. The method of any of Example Embodiments 1
to 23, further comprising:
[0172] receiving a training termination signal from the second
device; and
[0173] after receiving the training termination signal,
transitioning the first device from the training mode to a normal
operations mode.
Example Embodiment 25. The method of any of Example Embodiments 1
to 24, wherein the first device is user equipment and the second
device is a network device. Example Embodiment 26. A method in a
wireless communication network, the method comprising:
[0174] receiving, by a second device, information regarding an
artificial intelligence or machine learning (AI/ML) capability of a
first device over a single air interface between the first device
and the second device, the information regarding an AI/ML
capability of the first device identifying whether the first device
supports AI/ML for optimization of at least one air interface
configuration over the single air interface; and
[0175] transmitting an AI/ML training request to the first device
based at least in part on the information regarding the AI/ML
capability of the first device.
Example Embodiment 27. The method of Example Embodiment 26, wherein
the information regarding an AI/ML capability of the first device
comprises information indicating the first device is capable of
supporting a type and/or level of complexity of AI/ML. Example
Embodiment 28. The method of Example Embodiment 26 or 27, wherein
the information regarding an AI/ML capability of the first device
comprises information indicating whether the first device assists
with an AI/ML training process for optimization of the at least on
air interface configuration. Example Embodiment 29. The method of
any of Example Embodiments 26 to 28, wherein the information
regarding an AI/ML capability of the first device comprises
information indicating at least one component of the at least one
air interface configuration for which the first device supports
AI/ML optimization. Example Embodiment 30. The method of Example
Embodiment 29, wherein the at least one component of the at least
one air interface configuration includes at least one of a coding
component, a modulation component and a waveform component. Example
Embodiment 31. The method of Example Embodiment 29 or 30, wherein
the information indicating at least one component of the at least
one air interface configuration for which the first device supports
AI/ML optimization further comprises information indicating whether
the first device supports joint optimization of two or more
components of the at least one air interface configuration. Example
Embodiment 32. The method of any of Example Embodiments 26 to 31,
wherein receiving the information regarding an AI/ML capability of
the first device comprises receiving the information as part of an
initial network access procedure for the first device. Example
Embodiment 33. The method of any of Example Embodiments 26 to 32,
wherein transmitting the AI/ML training request comprises
transmitting the AI/ML training request through downlink control
information (DCI) on a downlink control channel or RRC signaling or
the combination of the DCI and RRC signaling. Example Embodiment
34. The method of Example Embodiment 33, further comprising,
receiving a training request response from the device confirming
that the device has transitioned to an AI/ML training mode. Example
Embodiment 35. The method of any of Example Embodiments 26 to 34,
further comprising transmitting a training signal to the first
device, the training signal including a training sequence or
training data for training at least one AI/ML module responsible
for one or more components of the at least one air interface
configuration. Example Embodiment 36. The method of Example
Embodiment 35, wherein transmitting the training signal comprises
transmitting the training signal on a dynamic control channel.
Example Embodiment 37. The method of Example Embodiment 36, wherein
the dynamic control channel includes a dynamic control information
(DCI) field containing information indicating an AI/ML module that
is to be trained. Example Embodiment 38. The method of Example
Embodiment 35, wherein transmitting the training signal comprises
transmitting the training signal on a scheduled data channel.
Example Embodiment 39. The method of Example Embodiment 38, further
comprising transmitting scheduling information for the data channel
on a dynamic control channel that includes a DCI field containing
information indicating an AI/ML module that is to be trained.
Example Embodiment 40. The method of any of Example Embodiments 35
to 39, further comprising receiving a training response message
from the first device, the training response message including
feedback information based on processing of the received training
signal at the first device. Example Embodiment 41. The method of
Example Embodiment 40, wherein the feedback information included in
the training response message includes an updated training sequence
for an iterative training process. Example Embodiment 42. The
method of Example Embodiment 40 or 41, wherein the feedback
information included in the training response message includes
measurement results based on the received training signal. Example
Embodiment 43. The method of Example Embodiment 42, wherein the
measurement results include an error margin obtained by the first
device in receiving the training signal. Example Embodiment 44. The
method of any of Example Embodiments 40 to 43, further
comprising:
[0176] training one or more AI/ML modules based on the feedback
information provided in the training response message from the
first device.
Example Embodiment 45. The method of Example Embodiment 44, further
comprising:
[0177] transmitting AI/ML update information to the first device,
the AI/ML update information including information indicating
updated AI/ML parameters for at least one of the one or more AI/ML
modules based on the training.
Example Embodiment 46. The method of any of Example Embodiments 40
to 45, further comprising:
[0178] receiving AI/ML update information from the first device,
the AI/ML update information from the first device including
information indicating updated AI/ML parameters for at least one of
the one or more AI/ML modules based on training of one or more
AI/ML modules at the first device based on the training signal.
Example Embodiment 47. The method of Example Embodiment 46, further
comprising updating the at least one air interface configuration
for transmitting to the first device by updating the one or more
AI/ML modules in accordance with the updated AI/ML parameters
transmitted to the first device and the updated AI/ML parameters
received from the first device. Example Embodiment 48. The method
of any of Example Embodiments 26 to 47, further comprising:
[0179] transmitting a training termination signal to the first
device to indicate that a training phase has finished.
Example Embodiment 49. The method of any of Example Embodiments 26
to 48, wherein the first device is user equipment and the second
device is a network device. Example Embodiment 50. An apparatus
comprising:
[0180] a wireless interface;
[0181] a processor operatively coupled to the wireless interface;
and
[0182] a computer readable storage medium operatively coupled to
the processor, the computer readable storage medium storing
programming for execution by the processor, the programming
comprising instructions to: [0183] transmit, from a first device
via the wireless interface, information regarding an artificial
intelligence or machine learning (AI/ML) capability of the first
device to a second device over a single air interface between the
first device and the second device, the information regarding an
AI/ML capability of the first device identifying whether the first
device supports AI/ML for optimization of at least one air
interface configuration over the single air interface. Example
Embodiment 51. The apparatus of Example Embodiment 50, wherein the
information regarding an AI/ML capability of the first device
comprises information indicating the first device is capable of
supporting a type and/or level of complexity of AI/ML. Example
Embodiment 52. The apparatus of Example Embodiment 50 or 51,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating whether the first device
assists with an AI/ML training process for optimization of the at
least one air interface configuration. Example Embodiment 53. The
apparatus of any of Example Embodiments 50 to 52, wherein the
information regarding an AI/ML capability of the first device
comprises information indicating at least one component of the at
least one air interface configuration for which the first device
supports AI/ML optimization. Example Embodiment 54. The apparatus
of Example Embodiment 53, wherein the at least one component of the
at least one air interface configuration includes at least one of a
coding component, a modulation component and a waveform component.
Example Embodiment 55. The apparatus of Example Embodiment 53 or
54, wherein the information indicating at least one component of
the at least one air interface configuration for which the first
device supports AI/ML optimization further comprises information
indicating whether the first device supports joint optimization of
two or more components of the at least one air interface
configuration. Example Embodiment 56. The apparatus of any of
Example Embodiments 50 to 55, wherein the instructions to transmit
the information regarding an AI/ML capability of the first device
comprises at least one of:
[0184] instructions to transmit the information in response to
receiving an enquiry; and
[0185] instructions to transmit the information as part of an
initial network access procedure.
Example Embodiment 57. The apparatus of any of Example Embodiments
50 to 56, wherein the programming further comprises instructions
to:
[0186] receive an AI/ML training request from the second device;
and
[0187] after receiving the AI/ML training request, transition to an
AI/ML training mode.
Example Embodiment 58. The apparatus of Example Embodiment 57,
wherein the instructions to receive the AI/ML training request
comprises instructions to receive the AI/ML training request
through downlink control information (DCI) on a downlink control
channel or RRC signaling or the combination of the DCI and RRC
signaling. Example Embodiment 59. The apparatus of Example
Embodiment 57 or 58, wherein the programming further comprises
instructions to transmit a training request response to the second
device to confirm that the first device has transitioned to the
AI/ML training mode. Example Embodiment 60. The apparatus of any of
Example Embodiments 50 to 59, wherein the programming further
comprises instructions to receive a training signal from the second
device that includes a training sequence or training data for
training at least one AI/ML module responsible for one or more
components of the at least one air interface configuration. Example
Embodiment 61. The apparatus of Example Embodiment 60, wherein the
instructions to receive the training signal comprise instructions
to receive the training signal on a dynamic control channel.
Example Embodiment 62. The apparatus of Example Embodiment 61,
wherein the dynamic control channel includes a dynamic control
information (DCI) field containing information indicating an AI/ML
module that is to be trained. Example Embodiment 63. The apparatus
of Example Embodiment 60, wherein the instructions to receive the
training signal comprise instructions to receive the training
signal on a scheduled data channel, the program further comprising
instructions to receive scheduling information for the data channel
on a dynamic control channel that includes a DCI field containing
information indicating an AI/ML module that is to be trained.
Example Embodiment 64. The apparatus of any of Example Embodiments
60 to 63, wherein the programming further comprises instructions
to:
[0188] transmit a training response message to the second device
after receiving the training signal, the training response message
including feedback information based on processing of the received
training signal at the first device.
Example Embodiment 65. The apparatus of Example Embodiment 64,
wherein the feedback information included in the training response
message includes an updated training sequence for an iterative
training process. Example Embodiment 66. The apparatus of Example
Embodiment 64 or 65, wherein the feedback information included in
the training response message includes measurement results based on
the received training signal. Example Embodiment 67. The apparatus
of Example Embodiment 66, wherein the measurement results include
an error margin obtained by the first device in receiving the
training signal from the second device. Example Embodiment 68. The
apparatus of any of Example Embodiments 64 to 67, wherein the
programming further comprises instructions to:
[0189] receive AI/ML update information from the second device
after transmitting the training response message, the AI/ML update
information including information indicating updated AI/ML
parameters for an AI/ML module based on the feedback information
provided by the first device.
Example Embodiment 69. The apparatus of Example Embodiment 68,
wherein the programming further comprises instructions to update
the AI/ML module in accordance with the updated AI/ML parameters in
order to update the at least one air interface configuration for
receiving transmissions from the second device. Example Embodiment
70. The apparatus of any of Example Embodiments 64 to 67, wherein
the programming further comprises instructions to:
[0190] train one or more AI/ML modules at the first device based on
the training signal received from the second device; and
[0191] transmit AI/ML update information to the second device, the
AI/ML update information including information indicating updated
AI/ML parameters for at least one of the one or more AI/ML modules
based on the training performed by the first device.
Example Embodiment 71. The apparatus of Example Embodiment 70,
wherein the programming further comprises instructions to receive
AI/ML update information from the second device, the AI/ML update
information from the second device including information indicating
updated AI/ML parameters for at least one of the one or more AI/ML
modules based on training of one or more AI/ML modules at the
second device based on feedback information provided in the
training response message. Example Embodiment 72. The apparatus of
Example Embodiment 71, wherein the programming further comprises
instructions to update the at least one air interface configuration
for receiving transmissions from the second device by updating the
one or more AI/ML modules in accordance with the updated AI/ML
parameters based on the training performed by the first device and
the updated AI/ML parameters received from the second device.
Example Embodiment 73. The apparatus of any of Example Embodiments
50 to 72, wherein the programming further comprises instructions
to:
[0192] receive a training termination signal from the second
device; and
[0193] after receiving the training termination signal, transition
the first device from the training mode to a normal operations
mode.
Example Embodiment 74. The apparatus of any of Example Embodiments
50 to 73, wherein the first device is user equipment and the second
device is a network device. Example Embodiment 75. An apparatus
comprising:
[0194] a wireless interface;
[0195] a processor operatively coupled to the wireless interface;
and
a computer readable storage medium operatively coupled to the
processor, the computer readable storage medium storing programming
for execution by the processor, the programming comprising
instructions to: [0196] receive, by a second device via the
wireless interface, information regarding an artificial
intelligence or machine learning (AI/ML) capability of a first
device over a single air interface between the first device and the
second device, the information regarding an AI/ML capability of the
first device identifying whether the first device supports AI/ML
for optimization of at least one air interface configuration over
the single air interface; and [0197] transmit an AI/ML training
request to the first device based at least in part on the
information regarding the AI/ML capability of the first device.
Example Embodiment 76. The apparatus of Example Embodiment 75,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating the first device is capable
of supporting a type and/or level of complexity of AI/ML. Example
Embodiment 77. The apparatus of Example Embodiment 75 or 76,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating whether the first device
assists with an AI/ML training process for optimization of the at
least on air interface configuration. Example Embodiment 78. The
apparatus of any of Example Embodiments 75 to 77, wherein the
information regarding an AI/ML capability of the first device
comprises information indicating at least one component of the at
least one air interface configuration for which the first device
supports AI/ML optimization. Example Embodiment 79. The apparatus
of Example Embodiment 78, wherein the at least one component of the
at least one air interface configuration includes at least one of a
coding component, a modulation component and a waveform component.
Example Embodiment 80. The apparatus of Example Embodiment 78 or
79, wherein the information indicating at least one component of
the at least one air interface configuration for which the first
device supports AI/ML optimization further comprises information
indicating whether the first device supports joint optimization of
two or more components of the at least one air interface
configuration. Example Embodiment 81. The apparatus of any of
Example Embodiments 75 to 80, wherein receiving the information
regarding an AI/ML capability of the first device comprises
receiving the information as part of an initial network access
procedure for the first device. Example Embodiment 82. The
apparatus of any of Example Embodiments 75 to 81, wherein
transmitting the AI/ML training request comprises transmitting the
AI/ML training request through downlink control information (DCI)
on a downlink control channel or RRC signaling or the combination
of the DCI and RRC signaling. Example Embodiment 83. The apparatus
of Example Embodiment 82, wherein the programming further comprises
instructions to receive a training request response from the device
confirming that the device has transitioned to an AI/ML training
mode. Example Embodiment 84. The apparatus of any of Example
Embodiments 75 to 83, wherein the programming further comprises
instructions to transmit a training signal to the first device, the
training signal including a training sequence or training data for
training at least one AI/ML module responsible for one or more
components of the at least one air interface configuration. Example
Embodiment 85. The apparatus of Example Embodiment 84, wherein
transmitting the training signal comprises transmitting the
training signal on a dynamic control channel. Example Embodiment
86. The apparatus of Example Embodiment 85, wherein the dynamic
control channel includes a dynamic control information (DCI) field
containing information indicating an AI/ML module that is to be
trained. Example Embodiment 87. The apparatus of Example Embodiment
84, wherein transmitting the training signal comprises transmitting
the training signal on a scheduled data channel. Example Embodiment
88. The apparatus of Example Embodiment 87, wherein the programming
further comprises instructions to transmit scheduling information
for the data channel on a dynamic control channel that includes a
DCI field containing information indicating an AI/ML module that is
to be trained. Example Embodiment 89. The apparatus of any of
Example Embodiments 84 to 88, wherein the programming further
comprises instructions to receive a training response message from
the first device, the training response message including feedback
information based on processing of the received training signal at
the first device. Example Embodiment 90. The apparatus of Example
Embodiment 89, wherein the feedback information included in the
training response message includes an updated training sequence for
an iterative training process. Example Embodiment 91. The apparatus
of Example Embodiment 89 or 90, wherein the feedback information
included in the training response message includes measurement
results based on the received training signal. Example Embodiment
92. The apparatus of Example Embodiment 91, wherein the measurement
results include an error margin obtained by the first device in
receiving the training signal. Example Embodiment 93. The apparatus
of any of Example Embodiments 89 to 92, wherein the programming
further comprises instructions to:
[0198] train one or more AI/ML modules based on the feedback
information provided in the training response message from the
first device.
Example Embodiment 94. The apparatus of Example Embodiment 93,
wherein the programming further comprises instructions to:
[0199] transmit AI/ML update information to the first device, the
AI/ML update information including information indicating updated
AI/ML parameters for at least one of the one or more AI/ML modules
based on the training.
Example Embodiment 95. The apparatus of any of Example Embodiments
89 to 94, wherein the programming further comprises instructions
to:
[0200] receive AI/ML update information from the first device, the
AI/ML update information from the first device including
information indicating updated AI/ML parameters for at least one of
the one or more AI/ML modules based on training of one or more
AI/ML modules at the first device based on the training signal.
Example Embodiment 96. The apparatus of Example Embodiment 95,
wherein the programming further comprises instructions to update
the at least one air interface configuration for transmitting to
the first device by updating the one or more AI/ML modules in
accordance with the updated AI/ML parameters transmitted to the
first device and the updated AI/ML parameters received from the
first device. Example Embodiment 97. The apparatus of any of
Example Embodiments 75 to 96, wherein the programming further
comprises instructions to:
[0201] transmit a training termination signal to the first device
to indicate that a training phase has finished.
Example Embodiment 98. The apparatus of any of Example Embodiments
75 to 97, wherein the first device is user equipment and the second
device is a network device. Example Embodiment 99. An apparatus
comprising:
[0202] a transmitting module configured to transmit, from a first
device, information regarding an artificial intelligence or machine
learning (AI/ML) capability of the first device to a second device
over an air interface between the first device and the second
device, the information regarding an AI/ML capability of the first
device identifying whether the first device supports AI/ML for
optimization of at least one air interface component over the air
interface.
Example Embodiment 100. The apparatus of Example Embodiment 99,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating the first device is capable
of supporting a type and/or level of complexity of AI/ML. Example
Embodiment 101. The apparatus of Example Embodiment 99 or 100,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating whether the first device
assists with an AI/ML training process for optimization of the at
least one air interface component. Example Embodiment 102. The
apparatus of any of Example Embodiments 99 to 101, wherein the
information regarding an AI/ML capability of the first device
comprises information indicating at least one component of the at
least one air interface component for which the first device
supports AI/ML optimization. Example Embodiment 103. The apparatus
of Example Embodiment 102, wherein the at least one air interface
component includes at least one of a coding component, a modulation
component and a waveform component. Example Embodiment 104. The
apparatus of Example Embodiment 102 or 103, wherein the information
indicating at least one component of the at least one air interface
component for which the first device supports AI/ML optimization
further comprises information indicating whether the first device
supports joint optimization of two or more components of the at
least one air interface component. Example Embodiment 105. The
apparatus of any of Example Embodiments 99 to 104, wherein the
transmitting module is configured to transmit the information
regarding an AI/ML capability of the first device in response to
receiving an enquiry or as part of an initial network access
procedure. Example Embodiment 106. The apparatus of any of Example
Embodiments 99 to 105, further comprising:
[0203] a receiving module configured to receive an AI/ML training
request from the second device; and
[0204] a processing module configured to transition to an AI/ML
training mode after the AI/ML training request is received.
Example Embodiment 107. The apparatus of Example Embodiment 106,
wherein the receiving module is configured to receive the AI/ML
training request through downlink control information (DCI) on a
downlink control channel or RRC signaling or the combination of the
DCI and RRC signaling. Example Embodiment 108. The apparatus of
Example Embodiment 106 or 107, wherein the transmitting module is
configured to transmit a training request response to the second
device to confirm that the first device has transitioned to the
AI/ML training mode. Example Embodiment 109. The apparatus of any
of Example Embodiments 99 to 108, wherein the receiving module is
configured to receive a training signal from the second device that
includes a training sequence or training data for training at least
one AI/ML module responsible for one or more components of the at
least one air interface component. Example Embodiment 110. The
apparatus of Example Embodiment 109, wherein the receiving module
is configured to receive the training signal on a dynamic control
channel. Example Embodiment 111. The apparatus of Example
Embodiment 110, wherein the dynamic control channel includes a
dynamic control information (DCI) field containing information
indicating an AI/ML module that is to be trained. Example
Embodiment 112. The apparatus of Example Embodiment 109, wherein
the receiving module is configured to:
[0205] receive the training signal on a scheduled data channel;
and
[0206] receive scheduling information for the data channel on a
dynamic control channel that includes a DCI field containing
information indicating an AI/ML module that is to be trained.
Example Embodiment 113. The apparatus of any of Example Embodiments
109 to 112, wherein the transmitting module is configured to:
[0207] transmit a training response message to the second device
after receiving the training signal, the training response message
including feedback information based on processing of the received
training signal at the first device.
Example Embodiment 114. The apparatus of Example Embodiment 113,
wherein the feedback information included in the training response
message includes an updated training sequence for an iterative
training process. Example Embodiment 115. The apparatus of Example
Embodiment 113 or 114, wherein the feedback information included in
the training response message includes measurement results based on
the received training signal. Example Embodiment 116. The apparatus
of Example Embodiment 115, wherein the measurement results include
an error margin obtained by the first device in receiving the
training signal from the second device. Example Embodiment 117. The
apparatus of any of Example Embodiments 113 to 116, wherein the
receiving module is configured to:
[0208] receive AI/ML update information from the second device
after transmitting the training response message, the AI/ML update
information including information indicating updated AI/ML
parameters for an AI/ML module based on the feedback information
provided by the first device.
Example Embodiment 118. The apparatus of Example Embodiment 117,
further comprising a processing module configured to update the
AI/ML module in accordance with the updated AI/ML parameters in
order to update the at least one air interface component for
receiving transmissions from the second device. Example Embodiment
119. The apparatus of any of Example Embodiments 113 to 116,
further comprising a processing module configured to train one or
more AI/ML modules at the first device based on the training signal
received from the second device, wherein the transmitting module is
configured to transmit AI/ML update information to the second
device, the AI/ML update information including information
indicating updated AI/ML parameters for at least one of the one or
more AI/ML modules based on the training performed by the first
device. Example Embodiment 120. The apparatus of Example Embodiment
119, wherein the receiving module is configured to receive AI/ML
update information from the second device, the AI/ML update
information from the second device including information indicating
updated AI/ML parameters for at least one of the one or more AI/ML
modules based on training of one or more AI/ML modules at the
second device based on feedback information provided in the
training response message. Example Embodiment 121. The apparatus of
Example Embodiment 120, wherein the processing module is configured
to update the at least one air interface component for receiving
transmissions from the second device by updating the one or more
AI/ML modules in accordance with the updated AI/ML parameters based
on the training performed by the first device and the updated AI/ML
parameters received from the second device. Example Embodiment 122.
The apparatus of any of Example Embodiments 99 to 121, wherein the
receiving module is configured to receive a training termination
signal from the second device, and the processing module is
configured to transition the first device from the training mode to
a normal operations mode after the training termination signal is
received. Example Embodiment 123. The apparatus of any of Example
Embodiments 99 to 122, wherein the first device is user equipment
and the second device is a network device. Example Embodiment 124.
An apparatus comprising:
[0209] a receiving module configured to receive, by a second
device, information regarding an artificial intelligence or machine
learning (AI/ML) capability of a first device over an air interface
between the first device and the second device, the information
regarding an AI/ML capability of the first device identifying
whether the first device supports AI/ML for optimization of at
least one air interface component over the air interface; and
[0210] a transmitting module configured to transmit an AI/ML
training request to the first device based at least in part on the
information regarding the AI/ML capability of the first device.
Example Embodiment 125. The apparatus of Example Embodiment 124,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating the first device is capable
of supporting a type and/or level of complexity of AI/ML. Example
Embodiment 126. The apparatus of Example Embodiment 124 or 125,
wherein the information regarding an AI/ML capability of the first
device comprises information indicating whether the first device
assists with an AI/ML training process for optimization of the at
least on air interface component. Example Embodiment 127. The
apparatus of any of Example Embodiments 124 to 126, wherein the
information regarding an AI/ML capability of the first device
comprises information indicating at least one component of the at
least one air interface component for which the first device
supports AI/ML optimization. Example Embodiment 128. The apparatus
of Example Embodiment 127, wherein the at least one component of
the at least one air interface component includes at least one of a
coding component, a modulation component and a waveform component.
Example Embodiment 129. The apparatus of Example Embodiment 127 or
128, wherein the information indicating at least one component of
the at least one air interface component for which the first device
supports AI/ML optimization further comprises information
indicating whether the first device supports joint optimization of
two or more components of the at least one air interface component.
Example Embodiment 130. The apparatus of any of Example Embodiments
124 to 129, wherein receiving the information regarding an AI/ML
capability of the first device comprises receiving the information
as part of an initial network access procedure for the first
device. Example Embodiment 131. The apparatus of any of Example
Embodiments 124 to 130, wherein transmitting the AI/ML training
request comprises transmitting the AI/ML training request through
downlink control information (DCI) on a downlink control channel or
RRC signaling or the combination of the DCI and RRC signaling.
Example Embodiment 132. The apparatus of Example Embodiment 131,
wherein the receiving module is configured to receive a training
request response from the device confirming that the device has
transitioned to an AI/ML training mode. Example Embodiment 133. The
apparatus of any of Example Embodiments 124 to 132, wherein the
transmitting module is configured to transmit a training signal to
the first device, the training signal including a training sequence
or training data for training at least one AI/ML module responsible
for one or more components of the at least one air interface
component. Example Embodiment 134. The apparatus of Example
Embodiment 133, wherein transmitting the training signal comprises
transmitting the training signal on a dynamic control channel.
Example Embodiment 135. The apparatus of Example Embodiment 134,
wherein the dynamic control channel includes a dynamic control
information (DCI) field containing information indicating an AI/ML
module that is to be trained. Example Embodiment 136. The apparatus
of Example Embodiment 133, wherein transmitting the training signal
comprises transmitting the training signal on a scheduled data
channel. Example Embodiment 137. The apparatus of Example
Embodiment 136, wherein the transmitting module is configured to
transmit scheduling information for the data channel on a dynamic
control channel that includes a DCI field containing information
indicating an AI/ML module that is to be trained. Example
Embodiment 138. The apparatus of any of Example Embodiments 133 to
137, wherein the receiving module is configured to receive a
training response message from the first device, the training
response message including feedback information based on processing
of the received training signal at the first device. Example
Embodiment 139. The apparatus of Example Embodiment 138, wherein
the feedback information included in the training response message
includes an updated training sequence for an iterative training
process. Example Embodiment 140. The apparatus of Example
Embodiment 138 or 139, wherein the feedback information included in
the training response message includes measurement results based on
the received training signal. Example Embodiment 141. The apparatus
of Example Embodiment 140, wherein the measurement results include
an error margin obtained by the first device in receiving the
training signal. Example Embodiment 142. The apparatus of any of
Example Embodiments 138 to 141, further comprising a processing
module configured to train one or more AI/ML modules based on the
feedback information provided in the training response message from
the first device. Example Embodiment 143. The apparatus of Example
Embodiment 142, wherein the transmitting module is configured
to:
[0211] transmit AI/ML update information to the first device, the
AI/ML update information including information indicating updated
AI/ML parameters for at least one of the one or more AI/ML modules
based on the training.
Example Embodiment 144. The apparatus of any of Example Embodiments
133 to 143, wherein the receiving module is configured to:
[0212] receive AI/ML update information from the first device, the
AI/ML update information from the first device including
information indicating updated AI/ML parameters for at least one of
the one or more AI/ML modules based on training of one or more
AI/ML modules at the first device based on the training signal.
Example Embodiment 145. The apparatus of Example Embodiment 144,
further comprising a processing module configured to update the at
least one air interface component for transmitting to the first
device by updating the one or more AI/ML modules in accordance with
the updated AI/ML parameters transmitted to the first device and
the updated AI/ML parameters received from the first device.
Example Embodiment 146. The apparatus of any of Example Embodiments
124 to 145, wherein the transmitting module is configured to
transmit a training termination signal to the first device to
indicate that a training phase has finished. Example Embodiment
147. The apparatus of any of Example Embodiments 124 to 146,
wherein the first device is user equipment and the second device is
a network device.
[0213] Although the present disclosure describes methods and
processes with steps in a certain order, one or more steps of the
methods and processes may be omitted or altered as appropriate. One
or more steps may take place in an order other than that in which
they are described, as appropriate.
[0214] Although the present disclosure is described, at least in
part, in terms of methods, a person of ordinary skill in the art
will understand that the present disclosure is also directed to the
various components for performing at least some of the aspects and
features of the described methods, be it by way of hardware
components, software or any combination of the two. Accordingly,
the technical solution of the present disclosure may be embodied in
the form of a software product. A suitable software product may be
stored in a pre-recorded storage device or other similar
non-volatile or non-transitory computer readable medium, including
DVDs, CD-ROMs, USB flash disk, a removable hard disk, or other
storage media, for example. The software product includes
instructions tangibly stored thereon that enable a processing
device (e.g., a personal computer, a server, or a network device)
to execute examples of the methods disclosed herein. The
machine-executable instructions may be in the form of code
sequences, configuration information, or other data, which, when
executed, cause a machine (e.g., a processor or other processing
device) to perform steps in a method according to examples of the
present disclosure.
[0215] The present disclosure may be embodied in other specific
forms without departing from the subject matter of the claims. The
described example embodiments are to be considered in all respects
as being only illustrative and not restrictive. Selected features
from one or more of the above-described embodiments may be combined
to create alternative embodiments not explicitly described,
features suitable for such combinations being understood within the
scope of this disclosure.
[0216] All values and sub-ranges within disclosed ranges are also
disclosed. Also, although the systems, devices and processes
disclosed and shown herein may comprise a specific number of
elements/components, the systems, devices and assemblies could be
modified to include additional or fewer of such
elements/components. For example, although any of the
elements/components disclosed may be referenced as being singular,
the embodiments disclosed herein could be modified to include a
plurality of such elements/components. The subject matter described
herein intends to cover and embrace all suitable changes in
technology.
* * * * *